From patchwork Mon Apr 5 14:00:51 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Baum X-Patchwork-Id: 90560 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 1B6BCA0A02; Mon, 5 Apr 2021 16:01:54 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id EA3A4140EED; Mon, 5 Apr 2021 16:01:38 +0200 (CEST) Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by mails.dpdk.org (Postfix) with ESMTP id 9442F140E95 for ; Mon, 5 Apr 2021 16:01:35 +0200 (CEST) Received: from Internal Mail-Server by MTLPINE1 (envelope-from michaelba@nvidia.com) with SMTP; 5 Apr 2021 17:01:32 +0300 Received: from nvidia.com (pegasus07.mtr.labs.mlnx [10.210.16.112]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 135E1VNX031476; Mon, 5 Apr 2021 17:01:31 +0300 From: Michael Baum To: dev@dpdk.org Cc: Matan Azrad , Raslan Darawsheh , Viacheslav Ovsiienko Date: Mon, 5 Apr 2021 14:00:51 +0000 Message-Id: <1617631256-3018-2-git-send-email-michaelba@nvidia.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> References: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> Subject: [dpdk-dev] [PATCH 1/6] net/mlx5: separate Rx function declarations to another file X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" The mlx5_rxtx.c file contains a lot of Tx burst functions, each of those is performance-optimized for the specific set of requested offloads. These ones are generated on the basis of the template function and it takes significant time to compile, just due to a large number of giant functions generated in the same file and this compilation is not being done in parallel with using multithreading. Therefore we can split the mlx5_rxtx.c file into several separate files to allow different functions to be compiled simultaneously. In this patch, we separate Rx function declarations to different header file in preparation for removing them from the source file and as an optional preparation step for further consolidation of Rx burst functions. Signed-off-by: Michael Baum Acked-by: Viacheslav Ovsiienko --- drivers/net/mlx5/linux/mlx5_mp_os.c | 1 + drivers/net/mlx5/linux/mlx5_os.c | 1 + drivers/net/mlx5/linux/mlx5_verbs.c | 1 + drivers/net/mlx5/mlx5.c | 1 + drivers/net/mlx5/mlx5_devx.c | 1 + drivers/net/mlx5/mlx5_ethdev.c | 1 + drivers/net/mlx5/mlx5_flow.c | 1 + drivers/net/mlx5/mlx5_flow_dv.c | 1 + drivers/net/mlx5/mlx5_flow_verbs.c | 1 + drivers/net/mlx5/mlx5_mr.c | 1 + drivers/net/mlx5/mlx5_rss.c | 1 + drivers/net/mlx5/mlx5_rx.h | 598 ++++++++++++++++++++++++++++++++++++ drivers/net/mlx5/mlx5_rxq.c | 1 + drivers/net/mlx5/mlx5_rxtx.c | 1 + drivers/net/mlx5/mlx5_rxtx.h | 569 ---------------------------------- drivers/net/mlx5/mlx5_rxtx_vec.c | 1 + drivers/net/mlx5/mlx5_stats.c | 1 + drivers/net/mlx5/mlx5_trigger.c | 1 + drivers/net/mlx5/mlx5_txpp.c | 1 + drivers/net/mlx5/mlx5_vlan.c | 1 + drivers/net/mlx5/windows/mlx5_os.c | 1 + 21 files changed, 617 insertions(+), 569 deletions(-) create mode 100644 drivers/net/mlx5/mlx5_rx.h diff --git a/drivers/net/mlx5/linux/mlx5_mp_os.c b/drivers/net/mlx5/linux/mlx5_mp_os.c index 8011ca8..63fa278 100644 --- a/drivers/net/mlx5/linux/mlx5_mp_os.c +++ b/drivers/net/mlx5/linux/mlx5_mp_os.c @@ -16,6 +16,7 @@ #include "mlx5.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_utils.h" int diff --git a/drivers/net/mlx5/linux/mlx5_os.c b/drivers/net/mlx5/linux/mlx5_os.c index 2d5bcab..97a28ec 100644 --- a/drivers/net/mlx5/linux/mlx5_os.c +++ b/drivers/net/mlx5/linux/mlx5_os.c @@ -40,6 +40,7 @@ #include "mlx5_common_os.h" #include "mlx5_utils.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" #include "mlx5_flow.h" diff --git a/drivers/net/mlx5/linux/mlx5_verbs.c b/drivers/net/mlx5/linux/mlx5_verbs.c index c7d4b17..73096af 100644 --- a/drivers/net/mlx5/linux/mlx5_verbs.c +++ b/drivers/net/mlx5/linux/mlx5_verbs.c @@ -22,6 +22,7 @@ #include #include #include +#include #include #include diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c index 9557d06..6f77bc2 100644 --- a/drivers/net/mlx5/mlx5.c +++ b/drivers/net/mlx5/mlx5.c @@ -35,6 +35,7 @@ #include "mlx5.h" #include "mlx5_utils.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" #include "mlx5_flow.h" diff --git a/drivers/net/mlx5/mlx5_devx.c b/drivers/net/mlx5/mlx5_devx.c index 5c940ed..76935f6 100644 --- a/drivers/net/mlx5/mlx5_devx.c +++ b/drivers/net/mlx5/mlx5_devx.c @@ -21,6 +21,7 @@ #include "mlx5.h" #include "mlx5_common_os.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_utils.h" #include "mlx5_devx.h" #include "mlx5_flow.h" diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c index 564d713..708e3a3 100644 --- a/drivers/net/mlx5/mlx5_ethdev.c +++ b/drivers/net/mlx5/mlx5_ethdev.c @@ -23,6 +23,7 @@ #include #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_autoconf.h" /** diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c index c347f81..b3877a1 100644 --- a/drivers/net/mlx5/mlx5_flow.c +++ b/drivers/net/mlx5/mlx5_flow.c @@ -30,6 +30,7 @@ #include "mlx5_flow.h" #include "mlx5_flow_os.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_common_os.h" #include "rte_pmd_mlx5.h" diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c index 533dadf..cac05fb 100644 --- a/drivers/net/mlx5/mlx5_flow_dv.c +++ b/drivers/net/mlx5/mlx5_flow_dv.c @@ -33,6 +33,7 @@ #include "mlx5_flow.h" #include "mlx5_flow_os.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "rte_pmd_mlx5.h" #if defined(HAVE_IBV_FLOW_DV_SUPPORT) || !defined(HAVE_INFINIBAND_VERBS_H) diff --git a/drivers/net/mlx5/mlx5_flow_verbs.c b/drivers/net/mlx5/mlx5_flow_verbs.c index b442b9b..c331350 100644 --- a/drivers/net/mlx5/mlx5_flow_verbs.c +++ b/drivers/net/mlx5/mlx5_flow_verbs.c @@ -24,6 +24,7 @@ #include "mlx5.h" #include "mlx5_flow.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #define VERBS_SPEC_INNER(item_flags) \ (!!((item_flags) & MLX5_FLOW_LAYER_TUNNEL) ? IBV_FLOW_SPEC_INNER : 0) diff --git a/drivers/net/mlx5/mlx5_mr.c b/drivers/net/mlx5/mlx5_mr.c index 3255393..2014936 100644 --- a/drivers/net/mlx5/mlx5_mr.c +++ b/drivers/net/mlx5/mlx5_mr.c @@ -15,6 +15,7 @@ #include "mlx5.h" #include "mlx5_mr.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" struct mr_find_contig_memsegs_data { uintptr_t addr; diff --git a/drivers/net/mlx5/mlx5_rss.c b/drivers/net/mlx5/mlx5_rss.c index dc0131a..c32129c 100644 --- a/drivers/net/mlx5/mlx5_rss.c +++ b/drivers/net/mlx5/mlx5_rss.c @@ -16,6 +16,7 @@ #include "mlx5_defs.h" #include "mlx5.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" /** * DPDK callback to update the RSS hash configuration. diff --git a/drivers/net/mlx5/mlx5_rx.h b/drivers/net/mlx5/mlx5_rx.h new file mode 100644 index 0000000..83b1f38 --- /dev/null +++ b/drivers/net/mlx5/mlx5_rx.h @@ -0,0 +1,598 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#ifndef RTE_PMD_MLX5_RX_H_ +#define RTE_PMD_MLX5_RX_H_ + +#include +#include + +#include +#include +#include +#include + +#include + +#include "mlx5.h" +#include "mlx5_autoconf.h" +#include "mlx5_mr.h" + +/* Support tunnel matching. */ +#define MLX5_FLOW_TUNNEL 10 + +struct mlx5_rxq_stats { +#ifdef MLX5_PMD_SOFT_COUNTERS + uint64_t ipackets; /**< Total of successfully received packets. */ + uint64_t ibytes; /**< Total of successfully received bytes. */ +#endif + uint64_t idropped; /**< Total of packets dropped when RX ring full. */ + uint64_t rx_nombuf; /**< Total of RX mbuf allocation failures. */ +}; + +/* Compressed CQE context. */ +struct rxq_zip { + uint16_t ai; /* Array index. */ + uint16_t ca; /* Current array index. */ + uint16_t na; /* Next array index. */ + uint16_t cq_ci; /* The next CQE. */ + uint32_t cqe_cnt; /* Number of CQEs. */ +}; + +/* Multi-Packet RQ buffer header. */ +struct mlx5_mprq_buf { + struct rte_mempool *mp; + uint16_t refcnt; /* Atomically accessed refcnt. */ + uint8_t pad[RTE_PKTMBUF_HEADROOM]; /* Headroom for the first packet. */ + struct rte_mbuf_ext_shared_info shinfos[]; + /* + * Shared information per stride. + * More memory will be allocated for the first stride head-room and for + * the strides data. + */ +} __rte_cache_aligned; + +/* Get pointer to the first stride. */ +#define mlx5_mprq_buf_addr(ptr, strd_n) (RTE_PTR_ADD((ptr), \ + sizeof(struct mlx5_mprq_buf) + \ + (strd_n) * \ + sizeof(struct rte_mbuf_ext_shared_info) + \ + RTE_PKTMBUF_HEADROOM)) + +#define MLX5_MIN_SINGLE_STRIDE_LOG_NUM_BYTES 6 +#define MLX5_MIN_SINGLE_WQE_LOG_NUM_STRIDES 9 + +enum mlx5_rxq_err_state { + MLX5_RXQ_ERR_STATE_NO_ERROR = 0, + MLX5_RXQ_ERR_STATE_NEED_RESET, + MLX5_RXQ_ERR_STATE_NEED_READY, +}; + +enum mlx5_rqx_code { + MLX5_RXQ_CODE_EXIT = 0, + MLX5_RXQ_CODE_NOMBUF, + MLX5_RXQ_CODE_DROPPED, +}; + +struct mlx5_eth_rxseg { + struct rte_mempool *mp; /**< Memory pool to allocate segment from. */ + uint16_t length; /**< Segment data length, configures split point. */ + uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */ + uint32_t reserved; /**< Reserved field. */ +}; + +/* RX queue descriptor. */ +struct mlx5_rxq_data { + unsigned int csum:1; /* Enable checksum offloading. */ + unsigned int hw_timestamp:1; /* Enable HW timestamp. */ + unsigned int rt_timestamp:1; /* Realtime timestamp format. */ + unsigned int vlan_strip:1; /* Enable VLAN stripping. */ + unsigned int crc_present:1; /* CRC must be subtracted. */ + unsigned int sges_n:3; /* Log 2 of SGEs (max buffers per packet). */ + unsigned int cqe_n:4; /* Log 2 of CQ elements. */ + unsigned int elts_n:4; /* Log 2 of Mbufs. */ + unsigned int rss_hash:1; /* RSS hash result is enabled. */ + unsigned int mark:1; /* Marked flow available on the queue. */ + unsigned int strd_num_n:5; /* Log 2 of the number of stride. */ + unsigned int strd_sz_n:4; /* Log 2 of stride size. */ + unsigned int strd_shift_en:1; /* Enable 2bytes shift on a stride. */ + unsigned int err_state:2; /* enum mlx5_rxq_err_state. */ + unsigned int strd_scatter_en:1; /* Scattered packets from a stride. */ + unsigned int lro:1; /* Enable LRO. */ + unsigned int dynf_meta:1; /* Dynamic metadata is configured. */ + unsigned int mcqe_format:3; /* CQE compression format. */ + volatile uint32_t *rq_db; + volatile uint32_t *cq_db; + uint16_t port_id; + uint32_t elts_ci; + uint32_t rq_ci; + uint16_t consumed_strd; /* Number of consumed strides in WQE. */ + uint32_t rq_pi; + uint32_t cq_ci; + uint16_t rq_repl_thresh; /* Threshold for buffer replenishment. */ + uint32_t byte_mask; + union { + struct rxq_zip zip; /* Compressed context. */ + uint16_t decompressed; + /* Number of ready mbufs decompressed from the CQ. */ + }; + struct mlx5_mr_ctrl mr_ctrl; /* MR control descriptor. */ + uint16_t mprq_max_memcpy_len; /* Maximum size of packet to memcpy. */ + volatile void *wqes; + volatile struct mlx5_cqe(*cqes)[]; + struct rte_mbuf *(*elts)[]; + struct mlx5_mprq_buf *(*mprq_bufs)[]; + struct rte_mempool *mp; + struct rte_mempool *mprq_mp; /* Mempool for Multi-Packet RQ. */ + struct mlx5_mprq_buf *mprq_repl; /* Stashed mbuf for replenish. */ + struct mlx5_dev_ctx_shared *sh; /* Shared context. */ + uint16_t idx; /* Queue index. */ + struct mlx5_rxq_stats stats; + rte_xmm_t mbuf_initializer; /* Default rearm/flags for vectorized Rx. */ + struct rte_mbuf fake_mbuf; /* elts padding for vectorized Rx. */ + void *cq_uar; /* Verbs CQ user access region. */ + uint32_t cqn; /* CQ number. */ + uint8_t cq_arm_sn; /* CQ arm seq number. */ +#ifndef RTE_ARCH_64 + rte_spinlock_t *uar_lock_cq; + /* CQ (UAR) access lock required for 32bit implementations */ +#endif + uint32_t tunnel; /* Tunnel information. */ + int timestamp_offset; /* Dynamic mbuf field for timestamp. */ + uint64_t timestamp_rx_flag; /* Dynamic mbuf flag for timestamp. */ + uint64_t flow_meta_mask; + int32_t flow_meta_offset; + uint32_t flow_meta_port_mask; + uint32_t rxseg_n; /* Number of split segment descriptions. */ + struct mlx5_eth_rxseg rxseg[MLX5_MAX_RXQ_NSEG]; + /* Buffer split segment descriptions - sizes, offsets, pools. */ +} __rte_cache_aligned; + +enum mlx5_rxq_type { + MLX5_RXQ_TYPE_STANDARD, /* Standard Rx queue. */ + MLX5_RXQ_TYPE_HAIRPIN, /* Hairpin Rx queue. */ + MLX5_RXQ_TYPE_UNDEFINED, +}; + +/* RX queue control descriptor. */ +struct mlx5_rxq_ctrl { + struct mlx5_rxq_data rxq; /* Data path structure. */ + LIST_ENTRY(mlx5_rxq_ctrl) next; /* Pointer to the next element. */ + uint32_t refcnt; /* Reference counter. */ + struct mlx5_rxq_obj *obj; /* Verbs/DevX elements. */ + struct mlx5_priv *priv; /* Back pointer to private data. */ + enum mlx5_rxq_type type; /* Rxq type. */ + unsigned int socket; /* CPU socket ID for allocations. */ + unsigned int irq:1; /* Whether IRQ is enabled. */ + uint32_t flow_mark_n; /* Number of Mark/Flag flows using this Queue. */ + uint32_t flow_tunnels_n[MLX5_FLOW_TUNNEL]; /* Tunnels counters. */ + uint32_t wqn; /* WQ number. */ + uint16_t dump_file_n; /* Number of dump files. */ + struct rte_eth_hairpin_conf hairpin_conf; /* Hairpin configuration. */ + uint32_t hairpin_status; /* Hairpin binding status. */ +}; + +/* mlx5_rxq.c */ + +extern uint8_t rss_hash_default_key[]; + +unsigned int mlx5_rxq_cqe_num(struct mlx5_rxq_data *rxq_data); +int mlx5_mprq_free_mp(struct rte_eth_dev *dev); +int mlx5_mprq_alloc_mp(struct rte_eth_dev *dev); +int mlx5_rx_queue_start(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_rx_queue_stop(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_rx_queue_start_primary(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_rx_queue_stop_primary(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_rx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, + unsigned int socket, const struct rte_eth_rxconf *conf, + struct rte_mempool *mp); +int mlx5_rx_hairpin_queue_setup + (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, + const struct rte_eth_hairpin_conf *hairpin_conf); +void mlx5_rx_queue_release(void *dpdk_rxq); +int mlx5_rx_intr_vec_enable(struct rte_eth_dev *dev); +void mlx5_rx_intr_vec_disable(struct rte_eth_dev *dev); +int mlx5_rx_intr_enable(struct rte_eth_dev *dev, uint16_t rx_queue_id); +int mlx5_rx_intr_disable(struct rte_eth_dev *dev, uint16_t rx_queue_id); +int mlx5_rxq_obj_verify(struct rte_eth_dev *dev); +struct mlx5_rxq_ctrl *mlx5_rxq_new(struct rte_eth_dev *dev, uint16_t idx, + uint16_t desc, unsigned int socket, + const struct rte_eth_rxconf *conf, + const struct rte_eth_rxseg_split *rx_seg, + uint16_t n_seg); +struct mlx5_rxq_ctrl *mlx5_rxq_hairpin_new + (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, + const struct rte_eth_hairpin_conf *hairpin_conf); +struct mlx5_rxq_ctrl *mlx5_rxq_get(struct rte_eth_dev *dev, uint16_t idx); +int mlx5_rxq_release(struct rte_eth_dev *dev, uint16_t idx); +int mlx5_rxq_verify(struct rte_eth_dev *dev); +int rxq_alloc_elts(struct mlx5_rxq_ctrl *rxq_ctrl); +int mlx5_ind_table_obj_verify(struct rte_eth_dev *dev); +struct mlx5_ind_table_obj *mlx5_ind_table_obj_get(struct rte_eth_dev *dev, + const uint16_t *queues, + uint32_t queues_n); +int mlx5_ind_table_obj_release(struct rte_eth_dev *dev, + struct mlx5_ind_table_obj *ind_tbl, + bool standalone); +int mlx5_ind_table_obj_setup(struct rte_eth_dev *dev, + struct mlx5_ind_table_obj *ind_tbl); +int mlx5_ind_table_obj_modify(struct rte_eth_dev *dev, + struct mlx5_ind_table_obj *ind_tbl, + uint16_t *queues, const uint32_t queues_n, + bool standalone); +struct mlx5_cache_entry *mlx5_hrxq_create_cb(struct mlx5_cache_list *list, + struct mlx5_cache_entry *entry __rte_unused, void *cb_ctx); +int mlx5_hrxq_match_cb(struct mlx5_cache_list *list, + struct mlx5_cache_entry *entry, + void *cb_ctx); +void mlx5_hrxq_remove_cb(struct mlx5_cache_list *list, + struct mlx5_cache_entry *entry); +uint32_t mlx5_hrxq_get(struct rte_eth_dev *dev, + struct mlx5_flow_rss_desc *rss_desc); +int mlx5_hrxq_release(struct rte_eth_dev *dev, uint32_t hxrq_idx); +uint32_t mlx5_hrxq_verify(struct rte_eth_dev *dev); +enum mlx5_rxq_type mlx5_rxq_get_type(struct rte_eth_dev *dev, uint16_t idx); +const struct rte_eth_hairpin_conf *mlx5_rxq_get_hairpin_conf + (struct rte_eth_dev *dev, uint16_t idx); +struct mlx5_hrxq *mlx5_drop_action_create(struct rte_eth_dev *dev); +void mlx5_drop_action_destroy(struct rte_eth_dev *dev); +uint64_t mlx5_get_rx_port_offloads(void); +uint64_t mlx5_get_rx_queue_offloads(struct rte_eth_dev *dev); +void mlx5_rxq_timestamp_set(struct rte_eth_dev *dev); +int mlx5_hrxq_modify(struct rte_eth_dev *dev, uint32_t hxrq_idx, + const uint8_t *rss_key, uint32_t rss_key_len, + uint64_t hash_fields, + const uint16_t *queues, uint32_t queues_n); + +/* mlx5_rxtx.c */ + +uint16_t mlx5_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n); +void mlx5_rxq_initialize(struct mlx5_rxq_data *rxq); +__rte_noinline int mlx5_rx_err_handle(struct mlx5_rxq_data *rxq, uint8_t vec); +void mlx5_mprq_buf_free_cb(void *addr, void *opaque); +void mlx5_mprq_buf_free(struct mlx5_mprq_buf *buf); +uint16_t mlx5_rx_burst_mprq(void *dpdk_rxq, struct rte_mbuf **pkts, + uint16_t pkts_n); +uint16_t removed_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, + uint16_t pkts_n); +int mlx5_rx_descriptor_status(void *rx_queue, uint16_t offset); +uint32_t mlx5_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id); +void mlx5_rxq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, + struct rte_eth_rxq_info *qinfo); +int mlx5_rx_burst_mode_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, + struct rte_eth_burst_mode *mode); + +/* Vectorized version of mlx5_rxtx.c */ +int mlx5_rxq_check_vec_support(struct mlx5_rxq_data *rxq_data); +int mlx5_check_vec_rx_support(struct rte_eth_dev *dev); +uint16_t mlx5_rx_burst_vec(void *dpdk_rxq, struct rte_mbuf **pkts, + uint16_t pkts_n); +uint16_t mlx5_rx_burst_mprq_vec(void *dpdk_rxq, struct rte_mbuf **pkts, + uint16_t pkts_n); + +/* mlx5_mr.c */ + +uint32_t mlx5_rx_addr2mr_bh(struct mlx5_rxq_data *rxq, uintptr_t addr); + +/** + * Query LKey from a packet buffer for Rx. No need to flush local caches for Rx + * as mempool is pre-configured and static. + * + * @param rxq + * Pointer to Rx queue structure. + * @param addr + * Address to search. + * + * @return + * Searched LKey on success, UINT32_MAX on no match. + */ +static __rte_always_inline uint32_t +mlx5_rx_addr2mr(struct mlx5_rxq_data *rxq, uintptr_t addr) +{ + struct mlx5_mr_ctrl *mr_ctrl = &rxq->mr_ctrl; + uint32_t lkey; + + /* Linear search on MR cache array. */ + lkey = mlx5_mr_lookup_lkey(mr_ctrl->cache, &mr_ctrl->mru, + MLX5_MR_CACHE_N, addr); + if (likely(lkey != UINT32_MAX)) + return lkey; + /* Take slower bottom-half (Binary Search) on miss. */ + return mlx5_rx_addr2mr_bh(rxq, addr); +} + +#define mlx5_rx_mb2mr(rxq, mb) mlx5_rx_addr2mr(rxq, (uintptr_t)((mb)->buf_addr)) + +/** + * Convert timestamp from HW format to linear counter + * from Packet Pacing Clock Queue CQE timestamp format. + * + * @param sh + * Pointer to the device shared context. Might be needed + * to convert according current device configuration. + * @param ts + * Timestamp from CQE to convert. + * @return + * UTC in nanoseconds + */ +static __rte_always_inline uint64_t +mlx5_txpp_convert_rx_ts(struct mlx5_dev_ctx_shared *sh, uint64_t ts) +{ + RTE_SET_USED(sh); + return (ts & UINT32_MAX) + (ts >> 32) * NS_PER_S; +} + +/** + * Set timestamp in mbuf dynamic field. + * + * @param mbuf + * Structure to write into. + * @param offset + * Dynamic field offset in mbuf structure. + * @param timestamp + * Value to write. + */ +static __rte_always_inline void +mlx5_timestamp_set(struct rte_mbuf *mbuf, int offset, + rte_mbuf_timestamp_t timestamp) +{ + *RTE_MBUF_DYNFIELD(mbuf, offset, rte_mbuf_timestamp_t *) = timestamp; +} + +/** + * Replace MPRQ buffer. + * + * @param rxq + * Pointer to Rx queue structure. + * @param rq_idx + * RQ index to replace. + */ +static __rte_always_inline void +mprq_buf_replace(struct mlx5_rxq_data *rxq, uint16_t rq_idx) +{ + const uint32_t strd_n = 1 << rxq->strd_num_n; + struct mlx5_mprq_buf *rep = rxq->mprq_repl; + volatile struct mlx5_wqe_data_seg *wqe = + &((volatile struct mlx5_wqe_mprq *)rxq->wqes)[rq_idx].dseg; + struct mlx5_mprq_buf *buf = (*rxq->mprq_bufs)[rq_idx]; + void *addr; + + if (__atomic_load_n(&buf->refcnt, __ATOMIC_RELAXED) > 1) { + MLX5_ASSERT(rep != NULL); + /* Replace MPRQ buf. */ + (*rxq->mprq_bufs)[rq_idx] = rep; + /* Replace WQE. */ + addr = mlx5_mprq_buf_addr(rep, strd_n); + wqe->addr = rte_cpu_to_be_64((uintptr_t)addr); + /* If there's only one MR, no need to replace LKey in WQE. */ + if (unlikely(mlx5_mr_btree_len(&rxq->mr_ctrl.cache_bh) > 1)) + wqe->lkey = mlx5_rx_addr2mr(rxq, (uintptr_t)addr); + /* Stash a mbuf for next replacement. */ + if (likely(!rte_mempool_get(rxq->mprq_mp, (void **)&rep))) + rxq->mprq_repl = rep; + else + rxq->mprq_repl = NULL; + /* Release the old buffer. */ + mlx5_mprq_buf_free(buf); + } else if (unlikely(rxq->mprq_repl == NULL)) { + struct mlx5_mprq_buf *rep; + + /* + * Currently, the MPRQ mempool is out of buffer + * and doing memcpy regardless of the size of Rx + * packet. Retry allocation to get back to + * normal. + */ + if (!rte_mempool_get(rxq->mprq_mp, (void **)&rep)) + rxq->mprq_repl = rep; + } +} + +/** + * Attach or copy MPRQ buffer content to a packet. + * + * @param rxq + * Pointer to Rx queue structure. + * @param pkt + * Pointer to a packet to fill. + * @param len + * Packet length. + * @param buf + * Pointer to a MPRQ buffer to take the data from. + * @param strd_idx + * Stride index to start from. + * @param strd_cnt + * Number of strides to consume. + */ +static __rte_always_inline enum mlx5_rqx_code +mprq_buf_to_pkt(struct mlx5_rxq_data *rxq, struct rte_mbuf *pkt, uint32_t len, + struct mlx5_mprq_buf *buf, uint16_t strd_idx, uint16_t strd_cnt) +{ + const uint32_t strd_n = 1 << rxq->strd_num_n; + const uint16_t strd_sz = 1 << rxq->strd_sz_n; + const uint16_t strd_shift = + MLX5_MPRQ_STRIDE_SHIFT_BYTE * rxq->strd_shift_en; + const int32_t hdrm_overlap = + len + RTE_PKTMBUF_HEADROOM - strd_cnt * strd_sz; + const uint32_t offset = strd_idx * strd_sz + strd_shift; + void *addr = RTE_PTR_ADD(mlx5_mprq_buf_addr(buf, strd_n), offset); + + /* + * Memcpy packets to the target mbuf if: + * - The size of packet is smaller than mprq_max_memcpy_len. + * - Out of buffer in the Mempool for Multi-Packet RQ. + * - The packet's stride overlaps a headroom and scatter is off. + */ + if (len <= rxq->mprq_max_memcpy_len || + rxq->mprq_repl == NULL || + (hdrm_overlap > 0 && !rxq->strd_scatter_en)) { + if (likely(len <= + (uint32_t)(pkt->buf_len - RTE_PKTMBUF_HEADROOM))) { + rte_memcpy(rte_pktmbuf_mtod(pkt, void *), + addr, len); + DATA_LEN(pkt) = len; + } else if (rxq->strd_scatter_en) { + struct rte_mbuf *prev = pkt; + uint32_t seg_len = RTE_MIN(len, (uint32_t) + (pkt->buf_len - RTE_PKTMBUF_HEADROOM)); + uint32_t rem_len = len - seg_len; + + rte_memcpy(rte_pktmbuf_mtod(pkt, void *), + addr, seg_len); + DATA_LEN(pkt) = seg_len; + while (rem_len) { + struct rte_mbuf *next = + rte_pktmbuf_alloc(rxq->mp); + + if (unlikely(next == NULL)) + return MLX5_RXQ_CODE_NOMBUF; + NEXT(prev) = next; + SET_DATA_OFF(next, 0); + addr = RTE_PTR_ADD(addr, seg_len); + seg_len = RTE_MIN(rem_len, (uint32_t) + (next->buf_len - RTE_PKTMBUF_HEADROOM)); + rte_memcpy + (rte_pktmbuf_mtod(next, void *), + addr, seg_len); + DATA_LEN(next) = seg_len; + rem_len -= seg_len; + prev = next; + ++NB_SEGS(pkt); + } + } else { + return MLX5_RXQ_CODE_DROPPED; + } + } else { + rte_iova_t buf_iova; + struct rte_mbuf_ext_shared_info *shinfo; + uint16_t buf_len = strd_cnt * strd_sz; + void *buf_addr; + + /* Increment the refcnt of the whole chunk. */ + __atomic_add_fetch(&buf->refcnt, 1, __ATOMIC_RELAXED); + MLX5_ASSERT(__atomic_load_n(&buf->refcnt, + __ATOMIC_RELAXED) <= strd_n + 1); + buf_addr = RTE_PTR_SUB(addr, RTE_PKTMBUF_HEADROOM); + /* + * MLX5 device doesn't use iova but it is necessary in a + * case where the Rx packet is transmitted via a + * different PMD. + */ + buf_iova = rte_mempool_virt2iova(buf) + + RTE_PTR_DIFF(buf_addr, buf); + shinfo = &buf->shinfos[strd_idx]; + rte_mbuf_ext_refcnt_set(shinfo, 1); + /* + * EXT_ATTACHED_MBUF will be set to pkt->ol_flags when + * attaching the stride to mbuf and more offload flags + * will be added below by calling rxq_cq_to_mbuf(). + * Other fields will be overwritten. + */ + rte_pktmbuf_attach_extbuf(pkt, buf_addr, buf_iova, + buf_len, shinfo); + /* Set mbuf head-room. */ + SET_DATA_OFF(pkt, RTE_PKTMBUF_HEADROOM); + MLX5_ASSERT(pkt->ol_flags == EXT_ATTACHED_MBUF); + MLX5_ASSERT(rte_pktmbuf_tailroom(pkt) >= + len - (hdrm_overlap > 0 ? hdrm_overlap : 0)); + DATA_LEN(pkt) = len; + /* + * Copy the last fragment of a packet (up to headroom + * size bytes) in case there is a stride overlap with + * a next packet's headroom. Allocate a separate mbuf + * to store this fragment and link it. Scatter is on. + */ + if (hdrm_overlap > 0) { + MLX5_ASSERT(rxq->strd_scatter_en); + struct rte_mbuf *seg = + rte_pktmbuf_alloc(rxq->mp); + + if (unlikely(seg == NULL)) + return MLX5_RXQ_CODE_NOMBUF; + SET_DATA_OFF(seg, 0); + rte_memcpy(rte_pktmbuf_mtod(seg, void *), + RTE_PTR_ADD(addr, len - hdrm_overlap), + hdrm_overlap); + DATA_LEN(seg) = hdrm_overlap; + DATA_LEN(pkt) = len - hdrm_overlap; + NEXT(pkt) = seg; + NB_SEGS(pkt) = 2; + } + } + return MLX5_RXQ_CODE_EXIT; +} + +/** + * Check whether Multi-Packet RQ can be enabled for the device. + * + * @param dev + * Pointer to Ethernet device. + * + * @return + * 1 if supported, negative errno value if not. + */ +static __rte_always_inline int +mlx5_check_mprq_support(struct rte_eth_dev *dev) +{ + struct mlx5_priv *priv = dev->data->dev_private; + + if (priv->config.mprq.enabled && + priv->rxqs_n >= priv->config.mprq.min_rxqs_num) + return 1; + return -ENOTSUP; +} + +/** + * Check whether Multi-Packet RQ is enabled for the Rx queue. + * + * @param rxq + * Pointer to receive queue structure. + * + * @return + * 0 if disabled, otherwise enabled. + */ +static __rte_always_inline int +mlx5_rxq_mprq_enabled(struct mlx5_rxq_data *rxq) +{ + return rxq->strd_num_n > 0; +} + +/** + * Check whether Multi-Packet RQ is enabled for the device. + * + * @param dev + * Pointer to Ethernet device. + * + * @return + * 0 if disabled, otherwise enabled. + */ +static __rte_always_inline int +mlx5_mprq_enabled(struct rte_eth_dev *dev) +{ + struct mlx5_priv *priv = dev->data->dev_private; + uint32_t i; + uint16_t n = 0; + uint16_t n_ibv = 0; + + if (mlx5_check_mprq_support(dev) < 0) + return 0; + /* All the configured queues should be enabled. */ + for (i = 0; i < priv->rxqs_n; ++i) { + struct mlx5_rxq_data *rxq = (*priv->rxqs)[i]; + struct mlx5_rxq_ctrl *rxq_ctrl = container_of + (rxq, struct mlx5_rxq_ctrl, rxq); + + if (rxq == NULL || rxq_ctrl->type != MLX5_RXQ_TYPE_STANDARD) + continue; + n_ibv++; + if (mlx5_rxq_mprq_enabled(rxq)) + ++n; + } + /* Multi-Packet RQ can't be partially configured. */ + MLX5_ASSERT(n == 0 || n == n_ibv); + return n == n_ibv; +} + +#endif /* RTE_PMD_MLX5_RX_H_ */ diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c index 9009eb8..19df0fa 100644 --- a/drivers/net/mlx5/mlx5_rxq.c +++ b/drivers/net/mlx5/mlx5_rxq.c @@ -25,6 +25,7 @@ #include "mlx5_defs.h" #include "mlx5.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_utils.h" #include "mlx5_autoconf.h" diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c index c76b995..d004e1e 100644 --- a/drivers/net/mlx5/mlx5_rxtx.c +++ b/drivers/net/mlx5/mlx5_rxtx.c @@ -25,6 +25,7 @@ #include "mlx5_mr.h" #include "mlx5_utils.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" /* TX burst subroutines return codes. */ enum mlx5_txcmp_code { diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h index 4f0fda0..d443db4 100644 --- a/drivers/net/mlx5/mlx5_rxtx.h +++ b/drivers/net/mlx5/mlx5_rxtx.h @@ -31,21 +31,10 @@ #include "mlx5_autoconf.h" #include "mlx5_mr.h" -/* Support tunnel matching. */ -#define MLX5_FLOW_TUNNEL 10 /* Mbuf dynamic flag offset for inline. */ extern uint64_t rte_net_mlx5_dynf_inline_mask; -struct mlx5_rxq_stats { -#ifdef MLX5_PMD_SOFT_COUNTERS - uint64_t ipackets; /**< Total of successfully received packets. */ - uint64_t ibytes; /**< Total of successfully received bytes. */ -#endif - uint64_t idropped; /**< Total of packets dropped when RX ring full. */ - uint64_t rx_nombuf; /**< Total of RX mbuf allocation failures. */ -}; - struct mlx5_txq_stats { #ifdef MLX5_PMD_SOFT_COUNTERS uint64_t opackets; /**< Total of successfully sent packets. */ @@ -56,148 +45,6 @@ struct mlx5_txq_stats { struct mlx5_priv; -/* Compressed CQE context. */ -struct rxq_zip { - uint16_t ai; /* Array index. */ - uint16_t ca; /* Current array index. */ - uint16_t na; /* Next array index. */ - uint16_t cq_ci; /* The next CQE. */ - uint32_t cqe_cnt; /* Number of CQEs. */ -}; - -/* Multi-Packet RQ buffer header. */ -struct mlx5_mprq_buf { - struct rte_mempool *mp; - uint16_t refcnt; /* Atomically accessed refcnt. */ - uint8_t pad[RTE_PKTMBUF_HEADROOM]; /* Headroom for the first packet. */ - struct rte_mbuf_ext_shared_info shinfos[]; - /* - * Shared information per stride. - * More memory will be allocated for the first stride head-room and for - * the strides data. - */ -} __rte_cache_aligned; - -/* Get pointer to the first stride. */ -#define mlx5_mprq_buf_addr(ptr, strd_n) (RTE_PTR_ADD((ptr), \ - sizeof(struct mlx5_mprq_buf) + \ - (strd_n) * \ - sizeof(struct rte_mbuf_ext_shared_info) + \ - RTE_PKTMBUF_HEADROOM)) - -#define MLX5_MIN_SINGLE_STRIDE_LOG_NUM_BYTES 6 -#define MLX5_MIN_SINGLE_WQE_LOG_NUM_STRIDES 9 - -enum mlx5_rxq_err_state { - MLX5_RXQ_ERR_STATE_NO_ERROR = 0, - MLX5_RXQ_ERR_STATE_NEED_RESET, - MLX5_RXQ_ERR_STATE_NEED_READY, -}; - -enum mlx5_rqx_code { - MLX5_RXQ_CODE_EXIT = 0, - MLX5_RXQ_CODE_NOMBUF, - MLX5_RXQ_CODE_DROPPED, -}; - -struct mlx5_eth_rxseg { - struct rte_mempool *mp; /**< Memory pool to allocate segment from. */ - uint16_t length; /**< Segment data length, configures split point. */ - uint16_t offset; /**< Data offset from beginning of mbuf data buffer. */ - uint32_t reserved; /**< Reserved field. */ -}; - -/* RX queue descriptor. */ -struct mlx5_rxq_data { - unsigned int csum:1; /* Enable checksum offloading. */ - unsigned int hw_timestamp:1; /* Enable HW timestamp. */ - unsigned int rt_timestamp:1; /* Realtime timestamp format. */ - unsigned int vlan_strip:1; /* Enable VLAN stripping. */ - unsigned int crc_present:1; /* CRC must be subtracted. */ - unsigned int sges_n:3; /* Log 2 of SGEs (max buffers per packet). */ - unsigned int cqe_n:4; /* Log 2 of CQ elements. */ - unsigned int elts_n:4; /* Log 2 of Mbufs. */ - unsigned int rss_hash:1; /* RSS hash result is enabled. */ - unsigned int mark:1; /* Marked flow available on the queue. */ - unsigned int strd_num_n:5; /* Log 2 of the number of stride. */ - unsigned int strd_sz_n:4; /* Log 2 of stride size. */ - unsigned int strd_shift_en:1; /* Enable 2bytes shift on a stride. */ - unsigned int err_state:2; /* enum mlx5_rxq_err_state. */ - unsigned int strd_scatter_en:1; /* Scattered packets from a stride. */ - unsigned int lro:1; /* Enable LRO. */ - unsigned int dynf_meta:1; /* Dynamic metadata is configured. */ - unsigned int mcqe_format:3; /* CQE compression format. */ - volatile uint32_t *rq_db; - volatile uint32_t *cq_db; - uint16_t port_id; - uint32_t elts_ci; - uint32_t rq_ci; - uint16_t consumed_strd; /* Number of consumed strides in WQE. */ - uint32_t rq_pi; - uint32_t cq_ci; - uint16_t rq_repl_thresh; /* Threshold for buffer replenishment. */ - uint32_t byte_mask; - union { - struct rxq_zip zip; /* Compressed context. */ - uint16_t decompressed; - /* Number of ready mbufs decompressed from the CQ. */ - }; - struct mlx5_mr_ctrl mr_ctrl; /* MR control descriptor. */ - uint16_t mprq_max_memcpy_len; /* Maximum size of packet to memcpy. */ - volatile void *wqes; - volatile struct mlx5_cqe(*cqes)[]; - struct rte_mbuf *(*elts)[]; - struct mlx5_mprq_buf *(*mprq_bufs)[]; - struct rte_mempool *mp; - struct rte_mempool *mprq_mp; /* Mempool for Multi-Packet RQ. */ - struct mlx5_mprq_buf *mprq_repl; /* Stashed mbuf for replenish. */ - struct mlx5_dev_ctx_shared *sh; /* Shared context. */ - uint16_t idx; /* Queue index. */ - struct mlx5_rxq_stats stats; - rte_xmm_t mbuf_initializer; /* Default rearm/flags for vectorized Rx. */ - struct rte_mbuf fake_mbuf; /* elts padding for vectorized Rx. */ - void *cq_uar; /* Verbs CQ user access region. */ - uint32_t cqn; /* CQ number. */ - uint8_t cq_arm_sn; /* CQ arm seq number. */ -#ifndef RTE_ARCH_64 - rte_spinlock_t *uar_lock_cq; - /* CQ (UAR) access lock required for 32bit implementations */ -#endif - uint32_t tunnel; /* Tunnel information. */ - int timestamp_offset; /* Dynamic mbuf field for timestamp. */ - uint64_t timestamp_rx_flag; /* Dynamic mbuf flag for timestamp. */ - uint64_t flow_meta_mask; - int32_t flow_meta_offset; - uint32_t flow_meta_port_mask; - uint32_t rxseg_n; /* Number of split segment descriptions. */ - struct mlx5_eth_rxseg rxseg[MLX5_MAX_RXQ_NSEG]; - /* Buffer split segment descriptions - sizes, offsets, pools. */ -} __rte_cache_aligned; - -enum mlx5_rxq_type { - MLX5_RXQ_TYPE_STANDARD, /* Standard Rx queue. */ - MLX5_RXQ_TYPE_HAIRPIN, /* Hairpin Rx queue. */ - MLX5_RXQ_TYPE_UNDEFINED, -}; - -/* RX queue control descriptor. */ -struct mlx5_rxq_ctrl { - struct mlx5_rxq_data rxq; /* Data path structure. */ - LIST_ENTRY(mlx5_rxq_ctrl) next; /* Pointer to the next element. */ - uint32_t refcnt; /* Reference counter. */ - struct mlx5_rxq_obj *obj; /* Verbs/DevX elements. */ - struct mlx5_priv *priv; /* Back pointer to private data. */ - enum mlx5_rxq_type type; /* Rxq type. */ - unsigned int socket; /* CPU socket ID for allocations. */ - unsigned int irq:1; /* Whether IRQ is enabled. */ - uint32_t flow_mark_n; /* Number of Mark/Flag flows using this Queue. */ - uint32_t flow_tunnels_n[MLX5_FLOW_TUNNEL]; /* Tunnels counters. */ - uint32_t wqn; /* WQ number. */ - uint16_t dump_file_n; /* Number of dump files. */ - struct rte_eth_hairpin_conf hairpin_conf; /* Hairpin configuration. */ - uint32_t hairpin_status; /* Hairpin binding status. */ -}; - /* TX queue send local data. */ __extension__ struct mlx5_txq_local { @@ -302,80 +149,6 @@ struct mlx5_txq_ctrl { #define MLX5_TX_BFREG(txq) \ (MLX5_PROC_PRIV((txq)->port_id)->uar_table[(txq)->idx]) -/* mlx5_rxq.c */ - -extern uint8_t rss_hash_default_key[]; - -unsigned int mlx5_rxq_cqe_num(struct mlx5_rxq_data *rxq_data); -int mlx5_mprq_free_mp(struct rte_eth_dev *dev); -int mlx5_mprq_alloc_mp(struct rte_eth_dev *dev); -int mlx5_rx_queue_start(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_rx_queue_stop(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_rx_queue_start_primary(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_rx_queue_stop_primary(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_rx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, - unsigned int socket, const struct rte_eth_rxconf *conf, - struct rte_mempool *mp); -int mlx5_rx_hairpin_queue_setup - (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, - const struct rte_eth_hairpin_conf *hairpin_conf); -void mlx5_rx_queue_release(void *dpdk_rxq); -int mlx5_rx_intr_vec_enable(struct rte_eth_dev *dev); -void mlx5_rx_intr_vec_disable(struct rte_eth_dev *dev); -int mlx5_rx_intr_enable(struct rte_eth_dev *dev, uint16_t rx_queue_id); -int mlx5_rx_intr_disable(struct rte_eth_dev *dev, uint16_t rx_queue_id); -int mlx5_rxq_obj_verify(struct rte_eth_dev *dev); -struct mlx5_rxq_ctrl *mlx5_rxq_new(struct rte_eth_dev *dev, uint16_t idx, - uint16_t desc, unsigned int socket, - const struct rte_eth_rxconf *conf, - const struct rte_eth_rxseg_split *rx_seg, - uint16_t n_seg); -struct mlx5_rxq_ctrl *mlx5_rxq_hairpin_new - (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, - const struct rte_eth_hairpin_conf *hairpin_conf); -struct mlx5_rxq_ctrl *mlx5_rxq_get(struct rte_eth_dev *dev, uint16_t idx); -int mlx5_rxq_release(struct rte_eth_dev *dev, uint16_t idx); -int mlx5_rxq_verify(struct rte_eth_dev *dev); -int rxq_alloc_elts(struct mlx5_rxq_ctrl *rxq_ctrl); -int mlx5_ind_table_obj_verify(struct rte_eth_dev *dev); -struct mlx5_ind_table_obj *mlx5_ind_table_obj_get(struct rte_eth_dev *dev, - const uint16_t *queues, - uint32_t queues_n); -int mlx5_ind_table_obj_release(struct rte_eth_dev *dev, - struct mlx5_ind_table_obj *ind_tbl, - bool standalone); -int mlx5_ind_table_obj_setup(struct rte_eth_dev *dev, - struct mlx5_ind_table_obj *ind_tbl); -int mlx5_ind_table_obj_modify(struct rte_eth_dev *dev, - struct mlx5_ind_table_obj *ind_tbl, - uint16_t *queues, const uint32_t queues_n, - bool standalone); -struct mlx5_cache_entry *mlx5_hrxq_create_cb(struct mlx5_cache_list *list, - struct mlx5_cache_entry *entry __rte_unused, void *cb_ctx); -int mlx5_hrxq_match_cb(struct mlx5_cache_list *list, - struct mlx5_cache_entry *entry, - void *cb_ctx); -void mlx5_hrxq_remove_cb(struct mlx5_cache_list *list, - struct mlx5_cache_entry *entry); -uint32_t mlx5_hrxq_get(struct rte_eth_dev *dev, - struct mlx5_flow_rss_desc *rss_desc); -int mlx5_hrxq_release(struct rte_eth_dev *dev, uint32_t hxrq_idx); -uint32_t mlx5_hrxq_verify(struct rte_eth_dev *dev); - - -enum mlx5_rxq_type mlx5_rxq_get_type(struct rte_eth_dev *dev, uint16_t idx); -const struct rte_eth_hairpin_conf *mlx5_rxq_get_hairpin_conf - (struct rte_eth_dev *dev, uint16_t idx); -struct mlx5_hrxq *mlx5_drop_action_create(struct rte_eth_dev *dev); -void mlx5_drop_action_destroy(struct rte_eth_dev *dev); -uint64_t mlx5_get_rx_port_offloads(void); -uint64_t mlx5_get_rx_queue_offloads(struct rte_eth_dev *dev); -void mlx5_rxq_timestamp_set(struct rte_eth_dev *dev); -int mlx5_hrxq_modify(struct rte_eth_dev *dev, uint32_t hxrq_idx, - const uint8_t *rss_key, uint32_t rss_key_len, - uint64_t hash_fields, - const uint16_t *queues, uint32_t queues_n); - /* mlx5_txq.c */ int mlx5_tx_queue_start(struct rte_eth_dev *dev, uint16_t queue_id); @@ -416,45 +189,21 @@ struct mlx5_txq_ctrl *mlx5_txq_hairpin_new void mlx5_set_ptype_table(void); void mlx5_set_cksum_table(void); void mlx5_set_swp_types_table(void); -uint16_t mlx5_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n); -void mlx5_rxq_initialize(struct mlx5_rxq_data *rxq); -__rte_noinline int mlx5_rx_err_handle(struct mlx5_rxq_data *rxq, uint8_t vec); -void mlx5_mprq_buf_free_cb(void *addr, void *opaque); -void mlx5_mprq_buf_free(struct mlx5_mprq_buf *buf); -uint16_t mlx5_rx_burst_mprq(void *dpdk_rxq, struct rte_mbuf **pkts, - uint16_t pkts_n); uint16_t removed_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n); -uint16_t removed_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, - uint16_t pkts_n); -int mlx5_rx_descriptor_status(void *rx_queue, uint16_t offset); int mlx5_tx_descriptor_status(void *tx_queue, uint16_t offset); -uint32_t mlx5_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id); void mlx5_dump_debug_information(const char *path, const char *title, const void *buf, unsigned int len); int mlx5_queue_state_modify_primary(struct rte_eth_dev *dev, const struct mlx5_mp_arg_queue_state_modify *sm); -void mlx5_rxq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, - struct rte_eth_rxq_info *qinfo); void mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, struct rte_eth_txq_info *qinfo); -int mlx5_rx_burst_mode_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, - struct rte_eth_burst_mode *mode); int mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, struct rte_eth_burst_mode *mode); -/* Vectorized version of mlx5_rxtx.c */ -int mlx5_rxq_check_vec_support(struct mlx5_rxq_data *rxq_data); -int mlx5_check_vec_rx_support(struct rte_eth_dev *dev); -uint16_t mlx5_rx_burst_vec(void *dpdk_txq, struct rte_mbuf **pkts, - uint16_t pkts_n); -uint16_t mlx5_rx_burst_mprq_vec(void *dpdk_txq, struct rte_mbuf **pkts, - uint16_t pkts_n); - /* mlx5_mr.c */ void mlx5_mr_flush_local_cache(struct mlx5_mr_ctrl *mr_ctrl); -uint32_t mlx5_rx_addr2mr_bh(struct mlx5_rxq_data *rxq, uintptr_t addr); uint32_t mlx5_tx_mb2mr_bh(struct mlx5_txq_data *txq, struct rte_mbuf *mb); uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, struct rte_mempool *mp); @@ -538,35 +287,6 @@ int mlx5_dma_unmap(struct rte_pci_device *pdev, void *addr, uint64_t iova, } /** - * Query LKey from a packet buffer for Rx. No need to flush local caches for Rx - * as mempool is pre-configured and static. - * - * @param rxq - * Pointer to Rx queue structure. - * @param addr - * Address to search. - * - * @return - * Searched LKey on success, UINT32_MAX on no match. - */ -static __rte_always_inline uint32_t -mlx5_rx_addr2mr(struct mlx5_rxq_data *rxq, uintptr_t addr) -{ - struct mlx5_mr_ctrl *mr_ctrl = &rxq->mr_ctrl; - uint32_t lkey; - - /* Linear search on MR cache array. */ - lkey = mlx5_mr_lookup_lkey(mr_ctrl->cache, &mr_ctrl->mru, - MLX5_MR_CACHE_N, addr); - if (likely(lkey != UINT32_MAX)) - return lkey; - /* Take slower bottom-half (Binary Search) on miss. */ - return mlx5_rx_addr2mr_bh(rxq, addr); -} - -#define mlx5_rx_mb2mr(rxq, mb) mlx5_rx_addr2mr(rxq, (uintptr_t)((mb)->buf_addr)) - -/** * Query LKey from a packet buffer for Tx. If not found, add the mempool. * * @param txq @@ -637,25 +357,6 @@ int mlx5_dma_unmap(struct rte_pci_device *pdev, void *addr, uint64_t iova, } /** - * Convert timestamp from HW format to linear counter - * from Packet Pacing Clock Queue CQE timestamp format. - * - * @param sh - * Pointer to the device shared context. Might be needed - * to convert according current device configuration. - * @param ts - * Timestamp from CQE to convert. - * @return - * UTC in nanoseconds - */ -static __rte_always_inline uint64_t -mlx5_txpp_convert_rx_ts(struct mlx5_dev_ctx_shared *sh, uint64_t ts) -{ - RTE_SET_USED(sh); - return (ts & UINT32_MAX) + (ts >> 32) * NS_PER_S; -} - -/** * Convert timestamp from mbuf format to linear counter * of Clock Queue completions (24 bits) * @@ -712,274 +413,4 @@ int mlx5_dma_unmap(struct rte_pci_device *pdev, void *addr, uint64_t iova, return ci; } -/** - * Set timestamp in mbuf dynamic field. - * - * @param mbuf - * Structure to write into. - * @param offset - * Dynamic field offset in mbuf structure. - * @param timestamp - * Value to write. - */ -static __rte_always_inline void -mlx5_timestamp_set(struct rte_mbuf *mbuf, int offset, - rte_mbuf_timestamp_t timestamp) -{ - *RTE_MBUF_DYNFIELD(mbuf, offset, rte_mbuf_timestamp_t *) = timestamp; -} - -/** - * Replace MPRQ buffer. - * - * @param rxq - * Pointer to Rx queue structure. - * @param rq_idx - * RQ index to replace. - */ -static __rte_always_inline void -mprq_buf_replace(struct mlx5_rxq_data *rxq, uint16_t rq_idx) -{ - const uint32_t strd_n = 1 << rxq->strd_num_n; - struct mlx5_mprq_buf *rep = rxq->mprq_repl; - volatile struct mlx5_wqe_data_seg *wqe = - &((volatile struct mlx5_wqe_mprq *)rxq->wqes)[rq_idx].dseg; - struct mlx5_mprq_buf *buf = (*rxq->mprq_bufs)[rq_idx]; - void *addr; - - if (__atomic_load_n(&buf->refcnt, __ATOMIC_RELAXED) > 1) { - MLX5_ASSERT(rep != NULL); - /* Replace MPRQ buf. */ - (*rxq->mprq_bufs)[rq_idx] = rep; - /* Replace WQE. */ - addr = mlx5_mprq_buf_addr(rep, strd_n); - wqe->addr = rte_cpu_to_be_64((uintptr_t)addr); - /* If there's only one MR, no need to replace LKey in WQE. */ - if (unlikely(mlx5_mr_btree_len(&rxq->mr_ctrl.cache_bh) > 1)) - wqe->lkey = mlx5_rx_addr2mr(rxq, (uintptr_t)addr); - /* Stash a mbuf for next replacement. */ - if (likely(!rte_mempool_get(rxq->mprq_mp, (void **)&rep))) - rxq->mprq_repl = rep; - else - rxq->mprq_repl = NULL; - /* Release the old buffer. */ - mlx5_mprq_buf_free(buf); - } else if (unlikely(rxq->mprq_repl == NULL)) { - struct mlx5_mprq_buf *rep; - - /* - * Currently, the MPRQ mempool is out of buffer - * and doing memcpy regardless of the size of Rx - * packet. Retry allocation to get back to - * normal. - */ - if (!rte_mempool_get(rxq->mprq_mp, (void **)&rep)) - rxq->mprq_repl = rep; - } -} - -/** - * Attach or copy MPRQ buffer content to a packet. - * - * @param rxq - * Pointer to Rx queue structure. - * @param pkt - * Pointer to a packet to fill. - * @param len - * Packet length. - * @param buf - * Pointer to a MPRQ buffer to take the data from. - * @param strd_idx - * Stride index to start from. - * @param strd_cnt - * Number of strides to consume. - */ -static __rte_always_inline enum mlx5_rqx_code -mprq_buf_to_pkt(struct mlx5_rxq_data *rxq, struct rte_mbuf *pkt, uint32_t len, - struct mlx5_mprq_buf *buf, uint16_t strd_idx, uint16_t strd_cnt) -{ - const uint32_t strd_n = 1 << rxq->strd_num_n; - const uint16_t strd_sz = 1 << rxq->strd_sz_n; - const uint16_t strd_shift = - MLX5_MPRQ_STRIDE_SHIFT_BYTE * rxq->strd_shift_en; - const int32_t hdrm_overlap = - len + RTE_PKTMBUF_HEADROOM - strd_cnt * strd_sz; - const uint32_t offset = strd_idx * strd_sz + strd_shift; - void *addr = RTE_PTR_ADD(mlx5_mprq_buf_addr(buf, strd_n), offset); - - /* - * Memcpy packets to the target mbuf if: - * - The size of packet is smaller than mprq_max_memcpy_len. - * - Out of buffer in the Mempool for Multi-Packet RQ. - * - The packet's stride overlaps a headroom and scatter is off. - */ - if (len <= rxq->mprq_max_memcpy_len || - rxq->mprq_repl == NULL || - (hdrm_overlap > 0 && !rxq->strd_scatter_en)) { - if (likely(len <= - (uint32_t)(pkt->buf_len - RTE_PKTMBUF_HEADROOM))) { - rte_memcpy(rte_pktmbuf_mtod(pkt, void *), - addr, len); - DATA_LEN(pkt) = len; - } else if (rxq->strd_scatter_en) { - struct rte_mbuf *prev = pkt; - uint32_t seg_len = RTE_MIN(len, (uint32_t) - (pkt->buf_len - RTE_PKTMBUF_HEADROOM)); - uint32_t rem_len = len - seg_len; - - rte_memcpy(rte_pktmbuf_mtod(pkt, void *), - addr, seg_len); - DATA_LEN(pkt) = seg_len; - while (rem_len) { - struct rte_mbuf *next = - rte_pktmbuf_alloc(rxq->mp); - - if (unlikely(next == NULL)) - return MLX5_RXQ_CODE_NOMBUF; - NEXT(prev) = next; - SET_DATA_OFF(next, 0); - addr = RTE_PTR_ADD(addr, seg_len); - seg_len = RTE_MIN(rem_len, (uint32_t) - (next->buf_len - RTE_PKTMBUF_HEADROOM)); - rte_memcpy - (rte_pktmbuf_mtod(next, void *), - addr, seg_len); - DATA_LEN(next) = seg_len; - rem_len -= seg_len; - prev = next; - ++NB_SEGS(pkt); - } - } else { - return MLX5_RXQ_CODE_DROPPED; - } - } else { - rte_iova_t buf_iova; - struct rte_mbuf_ext_shared_info *shinfo; - uint16_t buf_len = strd_cnt * strd_sz; - void *buf_addr; - - /* Increment the refcnt of the whole chunk. */ - __atomic_add_fetch(&buf->refcnt, 1, __ATOMIC_RELAXED); - MLX5_ASSERT(__atomic_load_n(&buf->refcnt, - __ATOMIC_RELAXED) <= strd_n + 1); - buf_addr = RTE_PTR_SUB(addr, RTE_PKTMBUF_HEADROOM); - /* - * MLX5 device doesn't use iova but it is necessary in a - * case where the Rx packet is transmitted via a - * different PMD. - */ - buf_iova = rte_mempool_virt2iova(buf) + - RTE_PTR_DIFF(buf_addr, buf); - shinfo = &buf->shinfos[strd_idx]; - rte_mbuf_ext_refcnt_set(shinfo, 1); - /* - * EXT_ATTACHED_MBUF will be set to pkt->ol_flags when - * attaching the stride to mbuf and more offload flags - * will be added below by calling rxq_cq_to_mbuf(). - * Other fields will be overwritten. - */ - rte_pktmbuf_attach_extbuf(pkt, buf_addr, buf_iova, - buf_len, shinfo); - /* Set mbuf head-room. */ - SET_DATA_OFF(pkt, RTE_PKTMBUF_HEADROOM); - MLX5_ASSERT(pkt->ol_flags == EXT_ATTACHED_MBUF); - MLX5_ASSERT(rte_pktmbuf_tailroom(pkt) >= - len - (hdrm_overlap > 0 ? hdrm_overlap : 0)); - DATA_LEN(pkt) = len; - /* - * Copy the last fragment of a packet (up to headroom - * size bytes) in case there is a stride overlap with - * a next packet's headroom. Allocate a separate mbuf - * to store this fragment and link it. Scatter is on. - */ - if (hdrm_overlap > 0) { - MLX5_ASSERT(rxq->strd_scatter_en); - struct rte_mbuf *seg = - rte_pktmbuf_alloc(rxq->mp); - - if (unlikely(seg == NULL)) - return MLX5_RXQ_CODE_NOMBUF; - SET_DATA_OFF(seg, 0); - rte_memcpy(rte_pktmbuf_mtod(seg, void *), - RTE_PTR_ADD(addr, len - hdrm_overlap), - hdrm_overlap); - DATA_LEN(seg) = hdrm_overlap; - DATA_LEN(pkt) = len - hdrm_overlap; - NEXT(pkt) = seg; - NB_SEGS(pkt) = 2; - } - } - return MLX5_RXQ_CODE_EXIT; -} - -/** - * Check whether Multi-Packet RQ can be enabled for the device. - * - * @param dev - * Pointer to Ethernet device. - * - * @return - * 1 if supported, negative errno value if not. - */ -static __rte_always_inline int -mlx5_check_mprq_support(struct rte_eth_dev *dev) -{ - struct mlx5_priv *priv = dev->data->dev_private; - - if (priv->config.mprq.enabled && - priv->rxqs_n >= priv->config.mprq.min_rxqs_num) - return 1; - return -ENOTSUP; -} - -/** - * Check whether Multi-Packet RQ is enabled for the Rx queue. - * - * @param rxq - * Pointer to receive queue structure. - * - * @return - * 0 if disabled, otherwise enabled. - */ -static __rte_always_inline int -mlx5_rxq_mprq_enabled(struct mlx5_rxq_data *rxq) -{ - return rxq->strd_num_n > 0; -} - -/** - * Check whether Multi-Packet RQ is enabled for the device. - * - * @param dev - * Pointer to Ethernet device. - * - * @return - * 0 if disabled, otherwise enabled. - */ -static __rte_always_inline int -mlx5_mprq_enabled(struct rte_eth_dev *dev) -{ - struct mlx5_priv *priv = dev->data->dev_private; - uint32_t i; - uint16_t n = 0; - uint16_t n_ibv = 0; - - if (mlx5_check_mprq_support(dev) < 0) - return 0; - /* All the configured queues should be enabled. */ - for (i = 0; i < priv->rxqs_n; ++i) { - struct mlx5_rxq_data *rxq = (*priv->rxqs)[i]; - struct mlx5_rxq_ctrl *rxq_ctrl = container_of - (rxq, struct mlx5_rxq_ctrl, rxq); - - if (rxq == NULL || rxq_ctrl->type != MLX5_RXQ_TYPE_STANDARD) - continue; - n_ibv++; - if (mlx5_rxq_mprq_enabled(rxq)) - ++n; - } - /* Multi-Packet RQ can't be partially configured. */ - MLX5_ASSERT(n == 0 || n == n_ibv); - return n == n_ibv; -} #endif /* RTE_PMD_MLX5_RXTX_H_ */ diff --git a/drivers/net/mlx5/mlx5_rxtx_vec.c b/drivers/net/mlx5/mlx5_rxtx_vec.c index 028e0f6..d5af2d9 100644 --- a/drivers/net/mlx5/mlx5_rxtx_vec.c +++ b/drivers/net/mlx5/mlx5_rxtx_vec.c @@ -19,6 +19,7 @@ #include "mlx5.h" #include "mlx5_utils.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_rxtx_vec.h" #include "mlx5_autoconf.h" diff --git a/drivers/net/mlx5/mlx5_stats.c b/drivers/net/mlx5/mlx5_stats.c index a6569b2..4dbd831 100644 --- a/drivers/net/mlx5/mlx5_stats.c +++ b/drivers/net/mlx5/mlx5_stats.c @@ -17,6 +17,7 @@ #include "mlx5_defs.h" #include "mlx5.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_malloc.h" /** diff --git a/drivers/net/mlx5/mlx5_trigger.c b/drivers/net/mlx5/mlx5_trigger.c index 94dd567..c88cb22 100644 --- a/drivers/net/mlx5/mlx5_trigger.c +++ b/drivers/net/mlx5/mlx5_trigger.c @@ -16,6 +16,7 @@ #include "mlx5.h" #include "mlx5_mr.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_utils.h" #include "rte_pmd_mlx5.h" diff --git a/drivers/net/mlx5/mlx5_txpp.c b/drivers/net/mlx5/mlx5_txpp.c index e8d632a..89e1c5d 100644 --- a/drivers/net/mlx5/mlx5_txpp.c +++ b/drivers/net/mlx5/mlx5_txpp.c @@ -17,6 +17,7 @@ #include "mlx5.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_common_os.h" static_assert(sizeof(struct mlx5_cqe_ts) == sizeof(rte_int128_t), diff --git a/drivers/net/mlx5/mlx5_vlan.c b/drivers/net/mlx5/mlx5_vlan.c index 64678d3..60f97f2 100644 --- a/drivers/net/mlx5/mlx5_vlan.c +++ b/drivers/net/mlx5/mlx5_vlan.c @@ -16,6 +16,7 @@ #include "mlx5.h" #include "mlx5_autoconf.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_utils.h" #include "mlx5_devx.h" diff --git a/drivers/net/mlx5/windows/mlx5_os.c b/drivers/net/mlx5/windows/mlx5_os.c index 6f39276..79eac80 100644 --- a/drivers/net/mlx5/windows/mlx5_os.c +++ b/drivers/net/mlx5/windows/mlx5_os.c @@ -23,6 +23,7 @@ #include "mlx5_common_os.h" #include "mlx5_utils.h" #include "mlx5_rxtx.h" +#include "mlx5_rx.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" #include "mlx5_flow.h" From patchwork Mon Apr 5 14:00:52 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Baum X-Patchwork-Id: 90562 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 4F7CAA0A02; Mon, 5 Apr 2021 16:02:14 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 6E4A7140F23; Mon, 5 Apr 2021 16:01:41 +0200 (CEST) Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by mails.dpdk.org (Postfix) with ESMTP id 9CD00140EA0 for ; Mon, 5 Apr 2021 16:01:35 +0200 (CEST) Received: from Internal Mail-Server by MTLPINE1 (envelope-from michaelba@nvidia.com) with SMTP; 5 Apr 2021 17:01:32 +0300 Received: from nvidia.com (pegasus07.mtr.labs.mlnx [10.210.16.112]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 135E1VNY031476; Mon, 5 Apr 2021 17:01:32 +0300 From: Michael Baum To: dev@dpdk.org Cc: Matan Azrad , Raslan Darawsheh , Viacheslav Ovsiienko Date: Mon, 5 Apr 2021 14:00:52 +0000 Message-Id: <1617631256-3018-3-git-send-email-michaelba@nvidia.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> References: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> Subject: [dpdk-dev] [PATCH 2/6] net/mlx5: separate Rx function implementations to new file X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" This patch separates Rx function implementations to different source file as an optional preparation step for further consolidation of Rx burst functions. Signed-off-by: Michael Baum Acked-by: Viacheslav Ovsiienko --- drivers/net/mlx5/meson.build | 1 + drivers/net/mlx5/mlx5_rx.c | 1203 ++++++++++++++++++++++++++++++++++++++++++ drivers/net/mlx5/mlx5_rx.h | 4 +- drivers/net/mlx5/mlx5_rxtx.c | 1179 +---------------------------------------- drivers/net/mlx5/mlx5_rxtx.h | 2 + 5 files changed, 1209 insertions(+), 1180 deletions(-) create mode 100644 drivers/net/mlx5/mlx5_rx.c diff --git a/drivers/net/mlx5/meson.build b/drivers/net/mlx5/meson.build index f2fafbd..0a89a27 100644 --- a/drivers/net/mlx5/meson.build +++ b/drivers/net/mlx5/meson.build @@ -19,6 +19,7 @@ sources = files( 'mlx5_mac.c', 'mlx5_mr.c', 'mlx5_rss.c', + 'mlx5_rx.c', 'mlx5_rxmode.c', 'mlx5_rxq.c', 'mlx5_rxtx.c', diff --git a/drivers/net/mlx5/mlx5_rx.c b/drivers/net/mlx5/mlx5_rx.c new file mode 100644 index 0000000..e9fcb52 --- /dev/null +++ b/drivers/net/mlx5/mlx5_rx.c @@ -0,0 +1,1203 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "mlx5_autoconf.h" +#include "mlx5_defs.h" +#include "mlx5.h" +#include "mlx5_mr.h" +#include "mlx5_utils.h" +#include "mlx5_rxtx.h" +#include "mlx5_rx.h" + + +static __rte_always_inline uint32_t +rxq_cq_to_pkt_type(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, + volatile struct mlx5_mini_cqe8 *mcqe); + +static __rte_always_inline int +mlx5_rx_poll_len(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, + uint16_t cqe_cnt, volatile struct mlx5_mini_cqe8 **mcqe); + +static __rte_always_inline uint32_t +rxq_cq_to_ol_flags(volatile struct mlx5_cqe *cqe); + +static __rte_always_inline void +rxq_cq_to_mbuf(struct mlx5_rxq_data *rxq, struct rte_mbuf *pkt, + volatile struct mlx5_cqe *cqe, + volatile struct mlx5_mini_cqe8 *mcqe); + +static inline void +mlx5_lro_update_tcp_hdr(struct rte_tcp_hdr *__rte_restrict tcp, + volatile struct mlx5_cqe *__rte_restrict cqe, + uint32_t phcsum, uint8_t l4_type); + +static inline void +mlx5_lro_update_hdr(uint8_t *__rte_restrict padd, + volatile struct mlx5_cqe *__rte_restrict cqe, + volatile struct mlx5_mini_cqe8 *mcqe, + struct mlx5_rxq_data *rxq, uint32_t len); + + +/** + * Internal function to compute the number of used descriptors in an RX queue. + * + * @param rxq + * The Rx queue. + * + * @return + * The number of used Rx descriptor. + */ +static uint32_t +rx_queue_count(struct mlx5_rxq_data *rxq) +{ + struct rxq_zip *zip = &rxq->zip; + volatile struct mlx5_cqe *cqe; + const unsigned int cqe_n = (1 << rxq->cqe_n); + const unsigned int sges_n = (1 << rxq->sges_n); + const unsigned int elts_n = (1 << rxq->elts_n); + const unsigned int strd_n = (1 << rxq->strd_num_n); + const unsigned int cqe_cnt = cqe_n - 1; + unsigned int cq_ci, used; + + /* if we are processing a compressed cqe */ + if (zip->ai) { + used = zip->cqe_cnt - zip->ai; + cq_ci = zip->cq_ci; + } else { + used = 0; + cq_ci = rxq->cq_ci; + } + cqe = &(*rxq->cqes)[cq_ci & cqe_cnt]; + while (check_cqe(cqe, cqe_n, cq_ci) != MLX5_CQE_STATUS_HW_OWN) { + int8_t op_own; + unsigned int n; + + op_own = cqe->op_own; + if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED) + n = rte_be_to_cpu_32(cqe->byte_cnt); + else + n = 1; + cq_ci += n; + used += n; + cqe = &(*rxq->cqes)[cq_ci & cqe_cnt]; + } + used = RTE_MIN(used * sges_n, elts_n * strd_n); + return used; +} + +/** + * DPDK callback to check the status of a Rx descriptor. + * + * @param rx_queue + * The Rx queue. + * @param[in] offset + * The index of the descriptor in the ring. + * + * @return + * The status of the Rx descriptor. + */ +int +mlx5_rx_descriptor_status(void *rx_queue, uint16_t offset) +{ + struct mlx5_rxq_data *rxq = rx_queue; + struct mlx5_rxq_ctrl *rxq_ctrl = + container_of(rxq, struct mlx5_rxq_ctrl, rxq); + struct rte_eth_dev *dev = ETH_DEV(rxq_ctrl->priv); + + if (dev->rx_pkt_burst == NULL || + dev->rx_pkt_burst == removed_rx_burst) { + rte_errno = ENOTSUP; + return -rte_errno; + } + if (offset >= (1 << rxq->cqe_n)) { + rte_errno = EINVAL; + return -rte_errno; + } + if (offset < rx_queue_count(rxq)) + return RTE_ETH_RX_DESC_DONE; + return RTE_ETH_RX_DESC_AVAIL; +} + +/** + * DPDK callback to get the RX queue information. + * + * @param dev + * Pointer to the device structure. + * + * @param rx_queue_id + * Rx queue identificator. + * + * @param qinfo + * Pointer to the RX queue information structure. + * + * @return + * None. + */ + +void +mlx5_rxq_info_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, + struct rte_eth_rxq_info *qinfo) +{ + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_rxq_data *rxq = (*priv->rxqs)[rx_queue_id]; + struct mlx5_rxq_ctrl *rxq_ctrl = + container_of(rxq, struct mlx5_rxq_ctrl, rxq); + + if (!rxq) + return; + qinfo->mp = mlx5_rxq_mprq_enabled(rxq) ? + rxq->mprq_mp : rxq->mp; + qinfo->conf.rx_thresh.pthresh = 0; + qinfo->conf.rx_thresh.hthresh = 0; + qinfo->conf.rx_thresh.wthresh = 0; + qinfo->conf.rx_free_thresh = rxq->rq_repl_thresh; + qinfo->conf.rx_drop_en = 1; + qinfo->conf.rx_deferred_start = rxq_ctrl ? 0 : 1; + qinfo->conf.offloads = dev->data->dev_conf.rxmode.offloads; + qinfo->scattered_rx = dev->data->scattered_rx; + qinfo->nb_desc = mlx5_rxq_mprq_enabled(rxq) ? + (1 << rxq->elts_n) * (1 << rxq->strd_num_n) : + (1 << rxq->elts_n); +} + +/** + * DPDK callback to get the RX packet burst mode information. + * + * @param dev + * Pointer to the device structure. + * + * @param rx_queue_id + * Rx queue identificatior. + * + * @param mode + * Pointer to the burts mode information. + * + * @return + * 0 as success, -EINVAL as failure. + */ +int +mlx5_rx_burst_mode_get(struct rte_eth_dev *dev, + uint16_t rx_queue_id __rte_unused, + struct rte_eth_burst_mode *mode) +{ + eth_rx_burst_t pkt_burst = dev->rx_pkt_burst; + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_rxq_data *rxq; + + rxq = (*priv->rxqs)[rx_queue_id]; + if (!rxq) { + rte_errno = EINVAL; + return -rte_errno; + } + if (pkt_burst == mlx5_rx_burst) { + snprintf(mode->info, sizeof(mode->info), "%s", "Scalar"); + } else if (pkt_burst == mlx5_rx_burst_mprq) { + snprintf(mode->info, sizeof(mode->info), "%s", "Multi-Packet RQ"); + } else if (pkt_burst == mlx5_rx_burst_vec) { +#if defined RTE_ARCH_X86_64 + snprintf(mode->info, sizeof(mode->info), "%s", "Vector SSE"); +#elif defined RTE_ARCH_ARM64 + snprintf(mode->info, sizeof(mode->info), "%s", "Vector Neon"); +#elif defined RTE_ARCH_PPC_64 + snprintf(mode->info, sizeof(mode->info), "%s", "Vector AltiVec"); +#else + return -EINVAL; +#endif + } else if (pkt_burst == mlx5_rx_burst_mprq_vec) { +#if defined RTE_ARCH_X86_64 + snprintf(mode->info, sizeof(mode->info), "%s", "MPRQ Vector SSE"); +#elif defined RTE_ARCH_ARM64 + snprintf(mode->info, sizeof(mode->info), "%s", "MPRQ Vector Neon"); +#elif defined RTE_ARCH_PPC_64 + snprintf(mode->info, sizeof(mode->info), "%s", "MPRQ Vector AltiVec"); +#else + return -EINVAL; +#endif + } else { + return -EINVAL; + } + return 0; +} + +/** + * DPDK callback to get the number of used descriptors in a RX queue. + * + * @param dev + * Pointer to the device structure. + * + * @param rx_queue_id + * The Rx queue. + * + * @return + * The number of used rx descriptor. + * -EINVAL if the queue is invalid + */ +uint32_t +mlx5_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id) +{ + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_rxq_data *rxq; + + if (dev->rx_pkt_burst == NULL || + dev->rx_pkt_burst == removed_rx_burst) { + rte_errno = ENOTSUP; + return -rte_errno; + } + rxq = (*priv->rxqs)[rx_queue_id]; + if (!rxq) { + rte_errno = EINVAL; + return -rte_errno; + } + return rx_queue_count(rxq); +} + +/** + * Translate RX completion flags to packet type. + * + * @param[in] rxq + * Pointer to RX queue structure. + * @param[in] cqe + * Pointer to CQE. + * + * @note: fix mlx5_dev_supported_ptypes_get() if any change here. + * + * @return + * Packet type for struct rte_mbuf. + */ +static inline uint32_t +rxq_cq_to_pkt_type(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, + volatile struct mlx5_mini_cqe8 *mcqe) +{ + uint8_t idx; + uint8_t ptype; + uint8_t pinfo = (cqe->pkt_info & 0x3) << 6; + + /* Get l3/l4 header from mini-CQE in case L3/L4 format*/ + if (mcqe == NULL || + rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_L34H_STRIDX) + ptype = (cqe->hdr_type_etc & 0xfc00) >> 10; + else + ptype = mcqe->hdr_type >> 2; + /* + * The index to the array should have: + * bit[1:0] = l3_hdr_type + * bit[4:2] = l4_hdr_type + * bit[5] = ip_frag + * bit[6] = tunneled + * bit[7] = outer_l3_type + */ + idx = pinfo | ptype; + return mlx5_ptype_table[idx] | rxq->tunnel * !!(idx & (1 << 6)); +} + +/** + * Initialize Rx WQ and indexes. + * + * @param[in] rxq + * Pointer to RX queue structure. + */ +void +mlx5_rxq_initialize(struct mlx5_rxq_data *rxq) +{ + const unsigned int wqe_n = 1 << rxq->elts_n; + unsigned int i; + + for (i = 0; (i != wqe_n); ++i) { + volatile struct mlx5_wqe_data_seg *scat; + uintptr_t addr; + uint32_t byte_count; + + if (mlx5_rxq_mprq_enabled(rxq)) { + struct mlx5_mprq_buf *buf = (*rxq->mprq_bufs)[i]; + + scat = &((volatile struct mlx5_wqe_mprq *) + rxq->wqes)[i].dseg; + addr = (uintptr_t)mlx5_mprq_buf_addr(buf, + 1 << rxq->strd_num_n); + byte_count = (1 << rxq->strd_sz_n) * + (1 << rxq->strd_num_n); + } else { + struct rte_mbuf *buf = (*rxq->elts)[i]; + + scat = &((volatile struct mlx5_wqe_data_seg *) + rxq->wqes)[i]; + addr = rte_pktmbuf_mtod(buf, uintptr_t); + byte_count = DATA_LEN(buf); + } + /* scat->addr must be able to store a pointer. */ + MLX5_ASSERT(sizeof(scat->addr) >= sizeof(uintptr_t)); + *scat = (struct mlx5_wqe_data_seg){ + .addr = rte_cpu_to_be_64(addr), + .byte_count = rte_cpu_to_be_32(byte_count), + .lkey = mlx5_rx_addr2mr(rxq, addr), + }; + } + rxq->consumed_strd = 0; + rxq->decompressed = 0; + rxq->rq_pi = 0; + rxq->zip = (struct rxq_zip){ + .ai = 0, + }; + rxq->elts_ci = mlx5_rxq_mprq_enabled(rxq) ? + (wqe_n >> rxq->sges_n) * (1 << rxq->strd_num_n) : 0; + /* Update doorbell counter. */ + rxq->rq_ci = wqe_n >> rxq->sges_n; + rte_io_wmb(); + *rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci); +} + +/** + * Handle a Rx error. + * The function inserts the RQ state to reset when the first error CQE is + * shown, then drains the CQ by the caller function loop. When the CQ is empty, + * it moves the RQ state to ready and initializes the RQ. + * Next CQE identification and error counting are in the caller responsibility. + * + * @param[in] rxq + * Pointer to RX queue structure. + * @param[in] vec + * 1 when called from vectorized Rx burst, need to prepare mbufs for the RQ. + * 0 when called from non-vectorized Rx burst. + * + * @return + * -1 in case of recovery error, otherwise the CQE status. + */ +int +mlx5_rx_err_handle(struct mlx5_rxq_data *rxq, uint8_t vec) +{ + const uint16_t cqe_n = 1 << rxq->cqe_n; + const uint16_t cqe_mask = cqe_n - 1; + const uint16_t wqe_n = 1 << rxq->elts_n; + const uint16_t strd_n = 1 << rxq->strd_num_n; + struct mlx5_rxq_ctrl *rxq_ctrl = + container_of(rxq, struct mlx5_rxq_ctrl, rxq); + union { + volatile struct mlx5_cqe *cqe; + volatile struct mlx5_err_cqe *err_cqe; + } u = { + .cqe = &(*rxq->cqes)[rxq->cq_ci & cqe_mask], + }; + struct mlx5_mp_arg_queue_state_modify sm; + int ret; + + switch (rxq->err_state) { + case MLX5_RXQ_ERR_STATE_NO_ERROR: + rxq->err_state = MLX5_RXQ_ERR_STATE_NEED_RESET; + /* Fall-through */ + case MLX5_RXQ_ERR_STATE_NEED_RESET: + sm.is_wq = 1; + sm.queue_id = rxq->idx; + sm.state = IBV_WQS_RESET; + if (mlx5_queue_state_modify(ETH_DEV(rxq_ctrl->priv), &sm)) + return -1; + if (rxq_ctrl->dump_file_n < + rxq_ctrl->priv->config.max_dump_files_num) { + MKSTR(err_str, "Unexpected CQE error syndrome " + "0x%02x CQN = %u RQN = %u wqe_counter = %u" + " rq_ci = %u cq_ci = %u", u.err_cqe->syndrome, + rxq->cqn, rxq_ctrl->wqn, + rte_be_to_cpu_16(u.err_cqe->wqe_counter), + rxq->rq_ci << rxq->sges_n, rxq->cq_ci); + MKSTR(name, "dpdk_mlx5_port_%u_rxq_%u_%u", + rxq->port_id, rxq->idx, (uint32_t)rte_rdtsc()); + mlx5_dump_debug_information(name, NULL, err_str, 0); + mlx5_dump_debug_information(name, "MLX5 Error CQ:", + (const void *)((uintptr_t) + rxq->cqes), + sizeof(*u.cqe) * cqe_n); + mlx5_dump_debug_information(name, "MLX5 Error RQ:", + (const void *)((uintptr_t) + rxq->wqes), + 16 * wqe_n); + rxq_ctrl->dump_file_n++; + } + rxq->err_state = MLX5_RXQ_ERR_STATE_NEED_READY; + /* Fall-through */ + case MLX5_RXQ_ERR_STATE_NEED_READY: + ret = check_cqe(u.cqe, cqe_n, rxq->cq_ci); + if (ret == MLX5_CQE_STATUS_HW_OWN) { + rte_io_wmb(); + *rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci); + rte_io_wmb(); + /* + * The RQ consumer index must be zeroed while moving + * from RESET state to RDY state. + */ + *rxq->rq_db = rte_cpu_to_be_32(0); + rte_io_wmb(); + sm.is_wq = 1; + sm.queue_id = rxq->idx; + sm.state = IBV_WQS_RDY; + if (mlx5_queue_state_modify(ETH_DEV(rxq_ctrl->priv), + &sm)) + return -1; + if (vec) { + const uint32_t elts_n = + mlx5_rxq_mprq_enabled(rxq) ? + wqe_n * strd_n : wqe_n; + const uint32_t e_mask = elts_n - 1; + uint32_t elts_ci = + mlx5_rxq_mprq_enabled(rxq) ? + rxq->elts_ci : rxq->rq_ci; + uint32_t elt_idx; + struct rte_mbuf **elt; + int i; + unsigned int n = elts_n - (elts_ci - + rxq->rq_pi); + + for (i = 0; i < (int)n; ++i) { + elt_idx = (elts_ci + i) & e_mask; + elt = &(*rxq->elts)[elt_idx]; + *elt = rte_mbuf_raw_alloc(rxq->mp); + if (!*elt) { + for (i--; i >= 0; --i) { + elt_idx = (elts_ci + + i) & elts_n; + elt = &(*rxq->elts) + [elt_idx]; + rte_pktmbuf_free_seg + (*elt); + } + return -1; + } + } + for (i = 0; i < (int)elts_n; ++i) { + elt = &(*rxq->elts)[i]; + DATA_LEN(*elt) = + (uint16_t)((*elt)->buf_len - + rte_pktmbuf_headroom(*elt)); + } + /* Padding with a fake mbuf for vec Rx. */ + for (i = 0; i < MLX5_VPMD_DESCS_PER_LOOP; ++i) + (*rxq->elts)[elts_n + i] = + &rxq->fake_mbuf; + } + mlx5_rxq_initialize(rxq); + rxq->err_state = MLX5_RXQ_ERR_STATE_NO_ERROR; + } + return ret; + default: + return -1; + } +} + +/** + * Get size of the next packet for a given CQE. For compressed CQEs, the + * consumer index is updated only once all packets of the current one have + * been processed. + * + * @param rxq + * Pointer to RX queue. + * @param cqe + * CQE to process. + * @param[out] mcqe + * Store pointer to mini-CQE if compressed. Otherwise, the pointer is not + * written. + * + * @return + * 0 in case of empty CQE, otherwise the packet size in bytes. + */ +static inline int +mlx5_rx_poll_len(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, + uint16_t cqe_cnt, volatile struct mlx5_mini_cqe8 **mcqe) +{ + struct rxq_zip *zip = &rxq->zip; + uint16_t cqe_n = cqe_cnt + 1; + int len; + uint16_t idx, end; + + do { + len = 0; + /* Process compressed data in the CQE and mini arrays. */ + if (zip->ai) { + volatile struct mlx5_mini_cqe8 (*mc)[8] = + (volatile struct mlx5_mini_cqe8 (*)[8]) + (uintptr_t)(&(*rxq->cqes)[zip->ca & + cqe_cnt].pkt_info); + len = rte_be_to_cpu_32((*mc)[zip->ai & 7].byte_cnt & + rxq->byte_mask); + *mcqe = &(*mc)[zip->ai & 7]; + if ((++zip->ai & 7) == 0) { + /* Invalidate consumed CQEs */ + idx = zip->ca; + end = zip->na; + while (idx != end) { + (*rxq->cqes)[idx & cqe_cnt].op_own = + MLX5_CQE_INVALIDATE; + ++idx; + } + /* + * Increment consumer index to skip the number + * of CQEs consumed. Hardware leaves holes in + * the CQ ring for software use. + */ + zip->ca = zip->na; + zip->na += 8; + } + if (unlikely(rxq->zip.ai == rxq->zip.cqe_cnt)) { + /* Invalidate the rest */ + idx = zip->ca; + end = zip->cq_ci; + + while (idx != end) { + (*rxq->cqes)[idx & cqe_cnt].op_own = + MLX5_CQE_INVALIDATE; + ++idx; + } + rxq->cq_ci = zip->cq_ci; + zip->ai = 0; + } + /* + * No compressed data, get next CQE and verify if it is + * compressed. + */ + } else { + int ret; + int8_t op_own; + uint32_t cq_ci; + + ret = check_cqe(cqe, cqe_n, rxq->cq_ci); + if (unlikely(ret != MLX5_CQE_STATUS_SW_OWN)) { + if (unlikely(ret == MLX5_CQE_STATUS_ERR || + rxq->err_state)) { + ret = mlx5_rx_err_handle(rxq, 0); + if (ret == MLX5_CQE_STATUS_HW_OWN || + ret == -1) + return 0; + } else { + return 0; + } + } + /* + * Introduce the local variable to have queue cq_ci + * index in queue structure always consistent with + * actual CQE boundary (not pointing to the middle + * of compressed CQE session). + */ + cq_ci = rxq->cq_ci + 1; + op_own = cqe->op_own; + if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED) { + volatile struct mlx5_mini_cqe8 (*mc)[8] = + (volatile struct mlx5_mini_cqe8 (*)[8]) + (uintptr_t)(&(*rxq->cqes) + [cq_ci & cqe_cnt].pkt_info); + + /* Fix endianness. */ + zip->cqe_cnt = rte_be_to_cpu_32(cqe->byte_cnt); + /* + * Current mini array position is the one + * returned by check_cqe64(). + * + * If completion comprises several mini arrays, + * as a special case the second one is located + * 7 CQEs after the initial CQE instead of 8 + * for subsequent ones. + */ + zip->ca = cq_ci; + zip->na = zip->ca + 7; + /* Compute the next non compressed CQE. */ + zip->cq_ci = rxq->cq_ci + zip->cqe_cnt; + /* Get packet size to return. */ + len = rte_be_to_cpu_32((*mc)[0].byte_cnt & + rxq->byte_mask); + *mcqe = &(*mc)[0]; + zip->ai = 1; + /* Prefetch all to be invalidated */ + idx = zip->ca; + end = zip->cq_ci; + while (idx != end) { + rte_prefetch0(&(*rxq->cqes)[(idx) & + cqe_cnt]); + ++idx; + } + } else { + rxq->cq_ci = cq_ci; + len = rte_be_to_cpu_32(cqe->byte_cnt); + } + } + if (unlikely(rxq->err_state)) { + cqe = &(*rxq->cqes)[rxq->cq_ci & cqe_cnt]; + ++rxq->stats.idropped; + } else { + return len; + } + } while (1); +} + +/** + * Translate RX completion flags to offload flags. + * + * @param[in] cqe + * Pointer to CQE. + * + * @return + * Offload flags (ol_flags) for struct rte_mbuf. + */ +static inline uint32_t +rxq_cq_to_ol_flags(volatile struct mlx5_cqe *cqe) +{ + uint32_t ol_flags = 0; + uint16_t flags = rte_be_to_cpu_16(cqe->hdr_type_etc); + + ol_flags = + TRANSPOSE(flags, + MLX5_CQE_RX_L3_HDR_VALID, + PKT_RX_IP_CKSUM_GOOD) | + TRANSPOSE(flags, + MLX5_CQE_RX_L4_HDR_VALID, + PKT_RX_L4_CKSUM_GOOD); + return ol_flags; +} + +/** + * Fill in mbuf fields from RX completion flags. + * Note that pkt->ol_flags should be initialized outside of this function. + * + * @param rxq + * Pointer to RX queue. + * @param pkt + * mbuf to fill. + * @param cqe + * CQE to process. + * @param rss_hash_res + * Packet RSS Hash result. + */ +static inline void +rxq_cq_to_mbuf(struct mlx5_rxq_data *rxq, struct rte_mbuf *pkt, + volatile struct mlx5_cqe *cqe, + volatile struct mlx5_mini_cqe8 *mcqe) +{ + /* Update packet information. */ + pkt->packet_type = rxq_cq_to_pkt_type(rxq, cqe, mcqe); + + if (rxq->rss_hash) { + uint32_t rss_hash_res = 0; + + /* If compressed, take hash result from mini-CQE. */ + if (mcqe == NULL || + rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_HASH) + rss_hash_res = rte_be_to_cpu_32(cqe->rx_hash_res); + else + rss_hash_res = rte_be_to_cpu_32(mcqe->rx_hash_result); + if (rss_hash_res) { + pkt->hash.rss = rss_hash_res; + pkt->ol_flags |= PKT_RX_RSS_HASH; + } + } + if (rxq->mark) { + uint32_t mark = 0; + + /* If compressed, take flow tag from mini-CQE. */ + if (mcqe == NULL || + rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_FTAG_STRIDX) + mark = cqe->sop_drop_qpn; + else + mark = ((mcqe->byte_cnt_flow & 0xff) << 8) | + (mcqe->flow_tag_high << 16); + if (MLX5_FLOW_MARK_IS_VALID(mark)) { + pkt->ol_flags |= PKT_RX_FDIR; + if (mark != RTE_BE32(MLX5_FLOW_MARK_DEFAULT)) { + pkt->ol_flags |= PKT_RX_FDIR_ID; + pkt->hash.fdir.hi = mlx5_flow_mark_get(mark); + } + } + } + if (rxq->dynf_meta) { + uint32_t meta = cqe->flow_table_metadata & + rxq->flow_meta_port_mask; + + if (meta) { + pkt->ol_flags |= rxq->flow_meta_mask; + *RTE_MBUF_DYNFIELD(pkt, rxq->flow_meta_offset, + uint32_t *) = meta; + } + } + if (rxq->csum) + pkt->ol_flags |= rxq_cq_to_ol_flags(cqe); + if (rxq->vlan_strip) { + bool vlan_strip; + + if (mcqe == NULL || + rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_L34H_STRIDX) + vlan_strip = cqe->hdr_type_etc & + RTE_BE16(MLX5_CQE_VLAN_STRIPPED); + else + vlan_strip = mcqe->hdr_type & + RTE_BE16(MLX5_CQE_VLAN_STRIPPED); + if (vlan_strip) { + pkt->ol_flags |= PKT_RX_VLAN | PKT_RX_VLAN_STRIPPED; + pkt->vlan_tci = rte_be_to_cpu_16(cqe->vlan_info); + } + } + if (rxq->hw_timestamp) { + uint64_t ts = rte_be_to_cpu_64(cqe->timestamp); + + if (rxq->rt_timestamp) + ts = mlx5_txpp_convert_rx_ts(rxq->sh, ts); + mlx5_timestamp_set(pkt, rxq->timestamp_offset, ts); + pkt->ol_flags |= rxq->timestamp_rx_flag; + } +} + +/** + * DPDK callback for RX. + * + * @param dpdk_rxq + * Generic pointer to RX queue structure. + * @param[out] pkts + * Array to store received packets. + * @param pkts_n + * Maximum number of packets in array. + * + * @return + * Number of packets successfully received (<= pkts_n). + */ +uint16_t +mlx5_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n) +{ + struct mlx5_rxq_data *rxq = dpdk_rxq; + const unsigned int wqe_cnt = (1 << rxq->elts_n) - 1; + const unsigned int cqe_cnt = (1 << rxq->cqe_n) - 1; + const unsigned int sges_n = rxq->sges_n; + struct rte_mbuf *pkt = NULL; + struct rte_mbuf *seg = NULL; + volatile struct mlx5_cqe *cqe = + &(*rxq->cqes)[rxq->cq_ci & cqe_cnt]; + unsigned int i = 0; + unsigned int rq_ci = rxq->rq_ci << sges_n; + int len = 0; /* keep its value across iterations. */ + + while (pkts_n) { + unsigned int idx = rq_ci & wqe_cnt; + volatile struct mlx5_wqe_data_seg *wqe = + &((volatile struct mlx5_wqe_data_seg *)rxq->wqes)[idx]; + struct rte_mbuf *rep = (*rxq->elts)[idx]; + volatile struct mlx5_mini_cqe8 *mcqe = NULL; + + if (pkt) + NEXT(seg) = rep; + seg = rep; + rte_prefetch0(seg); + rte_prefetch0(cqe); + rte_prefetch0(wqe); + /* Allocate the buf from the same pool. */ + rep = rte_mbuf_raw_alloc(seg->pool); + if (unlikely(rep == NULL)) { + ++rxq->stats.rx_nombuf; + if (!pkt) { + /* + * no buffers before we even started, + * bail out silently. + */ + break; + } + while (pkt != seg) { + MLX5_ASSERT(pkt != (*rxq->elts)[idx]); + rep = NEXT(pkt); + NEXT(pkt) = NULL; + NB_SEGS(pkt) = 1; + rte_mbuf_raw_free(pkt); + pkt = rep; + } + rq_ci >>= sges_n; + ++rq_ci; + rq_ci <<= sges_n; + break; + } + if (!pkt) { + cqe = &(*rxq->cqes)[rxq->cq_ci & cqe_cnt]; + len = mlx5_rx_poll_len(rxq, cqe, cqe_cnt, &mcqe); + if (!len) { + rte_mbuf_raw_free(rep); + break; + } + pkt = seg; + MLX5_ASSERT(len >= (rxq->crc_present << 2)); + pkt->ol_flags &= EXT_ATTACHED_MBUF; + rxq_cq_to_mbuf(rxq, pkt, cqe, mcqe); + if (rxq->crc_present) + len -= RTE_ETHER_CRC_LEN; + PKT_LEN(pkt) = len; + if (cqe->lro_num_seg > 1) { + mlx5_lro_update_hdr + (rte_pktmbuf_mtod(pkt, uint8_t *), cqe, + mcqe, rxq, len); + pkt->ol_flags |= PKT_RX_LRO; + pkt->tso_segsz = len / cqe->lro_num_seg; + } + } + DATA_LEN(rep) = DATA_LEN(seg); + PKT_LEN(rep) = PKT_LEN(seg); + SET_DATA_OFF(rep, DATA_OFF(seg)); + PORT(rep) = PORT(seg); + (*rxq->elts)[idx] = rep; + /* + * Fill NIC descriptor with the new buffer. The lkey and size + * of the buffers are already known, only the buffer address + * changes. + */ + wqe->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t)); + /* If there's only one MR, no need to replace LKey in WQE. */ + if (unlikely(mlx5_mr_btree_len(&rxq->mr_ctrl.cache_bh) > 1)) + wqe->lkey = mlx5_rx_mb2mr(rxq, rep); + if (len > DATA_LEN(seg)) { + len -= DATA_LEN(seg); + ++NB_SEGS(pkt); + ++rq_ci; + continue; + } + DATA_LEN(seg) = len; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Increment bytes counter. */ + rxq->stats.ibytes += PKT_LEN(pkt); +#endif + /* Return packet. */ + *(pkts++) = pkt; + pkt = NULL; + --pkts_n; + ++i; + /* Align consumer index to the next stride. */ + rq_ci >>= sges_n; + ++rq_ci; + rq_ci <<= sges_n; + } + if (unlikely(i == 0 && ((rq_ci >> sges_n) == rxq->rq_ci))) + return 0; + /* Update the consumer index. */ + rxq->rq_ci = rq_ci >> sges_n; + rte_io_wmb(); + *rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci); + rte_io_wmb(); + *rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci); +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Increment packets counter. */ + rxq->stats.ipackets += i; +#endif + return i; +} + +/** + * Update LRO packet TCP header. + * The HW LRO feature doesn't update the TCP header after coalescing the + * TCP segments but supplies information in CQE to fill it by SW. + * + * @param tcp + * Pointer to the TCP header. + * @param cqe + * Pointer to the completion entry. + * @param phcsum + * The L3 pseudo-header checksum. + */ +static inline void +mlx5_lro_update_tcp_hdr(struct rte_tcp_hdr *__rte_restrict tcp, + volatile struct mlx5_cqe *__rte_restrict cqe, + uint32_t phcsum, uint8_t l4_type) +{ + /* + * The HW calculates only the TCP payload checksum, need to complete + * the TCP header checksum and the L3 pseudo-header checksum. + */ + uint32_t csum = phcsum + cqe->csum; + + if (l4_type == MLX5_L4_HDR_TYPE_TCP_EMPTY_ACK || + l4_type == MLX5_L4_HDR_TYPE_TCP_WITH_ACL) { + tcp->tcp_flags |= RTE_TCP_ACK_FLAG; + tcp->recv_ack = cqe->lro_ack_seq_num; + tcp->rx_win = cqe->lro_tcp_win; + } + if (cqe->lro_tcppsh_abort_dupack & MLX5_CQE_LRO_PUSH_MASK) + tcp->tcp_flags |= RTE_TCP_PSH_FLAG; + tcp->cksum = 0; + csum += rte_raw_cksum(tcp, (tcp->data_off >> 4) * 4); + csum = ((csum & 0xffff0000) >> 16) + (csum & 0xffff); + csum = (~csum) & 0xffff; + if (csum == 0) + csum = 0xffff; + tcp->cksum = csum; +} + +/** + * Update LRO packet headers. + * The HW LRO feature doesn't update the L3/TCP headers after coalescing the + * TCP segments but supply information in CQE to fill it by SW. + * + * @param padd + * The packet address. + * @param cqe + * Pointer to the completion entry. + * @param len + * The packet length. + */ +static inline void +mlx5_lro_update_hdr(uint8_t *__rte_restrict padd, + volatile struct mlx5_cqe *__rte_restrict cqe, + volatile struct mlx5_mini_cqe8 *mcqe, + struct mlx5_rxq_data *rxq, uint32_t len) +{ + union { + struct rte_ether_hdr *eth; + struct rte_vlan_hdr *vlan; + struct rte_ipv4_hdr *ipv4; + struct rte_ipv6_hdr *ipv6; + struct rte_tcp_hdr *tcp; + uint8_t *hdr; + } h = { + .hdr = padd, + }; + uint16_t proto = h.eth->ether_type; + uint32_t phcsum; + uint8_t l4_type; + + h.eth++; + while (proto == RTE_BE16(RTE_ETHER_TYPE_VLAN) || + proto == RTE_BE16(RTE_ETHER_TYPE_QINQ)) { + proto = h.vlan->eth_proto; + h.vlan++; + } + if (proto == RTE_BE16(RTE_ETHER_TYPE_IPV4)) { + h.ipv4->time_to_live = cqe->lro_min_ttl; + h.ipv4->total_length = rte_cpu_to_be_16(len - (h.hdr - padd)); + h.ipv4->hdr_checksum = 0; + h.ipv4->hdr_checksum = rte_ipv4_cksum(h.ipv4); + phcsum = rte_ipv4_phdr_cksum(h.ipv4, 0); + h.ipv4++; + } else { + h.ipv6->hop_limits = cqe->lro_min_ttl; + h.ipv6->payload_len = rte_cpu_to_be_16(len - (h.hdr - padd) - + sizeof(*h.ipv6)); + phcsum = rte_ipv6_phdr_cksum(h.ipv6, 0); + h.ipv6++; + } + if (mcqe == NULL || + rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_L34H_STRIDX) + l4_type = (rte_be_to_cpu_16(cqe->hdr_type_etc) & + MLX5_CQE_L4_TYPE_MASK) >> MLX5_CQE_L4_TYPE_SHIFT; + else + l4_type = (rte_be_to_cpu_16(mcqe->hdr_type) & + MLX5_CQE_L4_TYPE_MASK) >> MLX5_CQE_L4_TYPE_SHIFT; + mlx5_lro_update_tcp_hdr(h.tcp, cqe, phcsum, l4_type); +} + +void +mlx5_mprq_buf_free_cb(void *addr __rte_unused, void *opaque) +{ + struct mlx5_mprq_buf *buf = opaque; + + if (__atomic_load_n(&buf->refcnt, __ATOMIC_RELAXED) == 1) { + rte_mempool_put(buf->mp, buf); + } else if (unlikely(__atomic_sub_fetch(&buf->refcnt, 1, + __ATOMIC_RELAXED) == 0)) { + __atomic_store_n(&buf->refcnt, 1, __ATOMIC_RELAXED); + rte_mempool_put(buf->mp, buf); + } +} + +void +mlx5_mprq_buf_free(struct mlx5_mprq_buf *buf) +{ + mlx5_mprq_buf_free_cb(NULL, buf); +} + +/** + * DPDK callback for RX with Multi-Packet RQ support. + * + * @param dpdk_rxq + * Generic pointer to RX queue structure. + * @param[out] pkts + * Array to store received packets. + * @param pkts_n + * Maximum number of packets in array. + * + * @return + * Number of packets successfully received (<= pkts_n). + */ +uint16_t +mlx5_rx_burst_mprq(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n) +{ + struct mlx5_rxq_data *rxq = dpdk_rxq; + const uint32_t strd_n = 1 << rxq->strd_num_n; + const uint32_t strd_sz = 1 << rxq->strd_sz_n; + const uint32_t cq_mask = (1 << rxq->cqe_n) - 1; + const uint32_t wq_mask = (1 << rxq->elts_n) - 1; + volatile struct mlx5_cqe *cqe = &(*rxq->cqes)[rxq->cq_ci & cq_mask]; + unsigned int i = 0; + uint32_t rq_ci = rxq->rq_ci; + uint16_t consumed_strd = rxq->consumed_strd; + struct mlx5_mprq_buf *buf = (*rxq->mprq_bufs)[rq_ci & wq_mask]; + + while (i < pkts_n) { + struct rte_mbuf *pkt; + int ret; + uint32_t len; + uint16_t strd_cnt; + uint16_t strd_idx; + uint32_t byte_cnt; + volatile struct mlx5_mini_cqe8 *mcqe = NULL; + enum mlx5_rqx_code rxq_code; + + if (consumed_strd == strd_n) { + /* Replace WQE if the buffer is still in use. */ + mprq_buf_replace(rxq, rq_ci & wq_mask); + /* Advance to the next WQE. */ + consumed_strd = 0; + ++rq_ci; + buf = (*rxq->mprq_bufs)[rq_ci & wq_mask]; + } + cqe = &(*rxq->cqes)[rxq->cq_ci & cq_mask]; + ret = mlx5_rx_poll_len(rxq, cqe, cq_mask, &mcqe); + if (!ret) + break; + byte_cnt = ret; + len = (byte_cnt & MLX5_MPRQ_LEN_MASK) >> MLX5_MPRQ_LEN_SHIFT; + MLX5_ASSERT((int)len >= (rxq->crc_present << 2)); + if (rxq->crc_present) + len -= RTE_ETHER_CRC_LEN; + if (mcqe && + rxq->mcqe_format == MLX5_CQE_RESP_FORMAT_FTAG_STRIDX) + strd_cnt = (len / strd_sz) + !!(len % strd_sz); + else + strd_cnt = (byte_cnt & MLX5_MPRQ_STRIDE_NUM_MASK) >> + MLX5_MPRQ_STRIDE_NUM_SHIFT; + MLX5_ASSERT(strd_cnt); + consumed_strd += strd_cnt; + if (byte_cnt & MLX5_MPRQ_FILLER_MASK) + continue; + strd_idx = rte_be_to_cpu_16(mcqe == NULL ? + cqe->wqe_counter : + mcqe->stride_idx); + MLX5_ASSERT(strd_idx < strd_n); + MLX5_ASSERT(!((rte_be_to_cpu_16(cqe->wqe_id) ^ rq_ci) & + wq_mask)); + pkt = rte_pktmbuf_alloc(rxq->mp); + if (unlikely(pkt == NULL)) { + ++rxq->stats.rx_nombuf; + break; + } + len = (byte_cnt & MLX5_MPRQ_LEN_MASK) >> MLX5_MPRQ_LEN_SHIFT; + MLX5_ASSERT((int)len >= (rxq->crc_present << 2)); + if (rxq->crc_present) + len -= RTE_ETHER_CRC_LEN; + rxq_code = mprq_buf_to_pkt(rxq, pkt, len, buf, + strd_idx, strd_cnt); + if (unlikely(rxq_code != MLX5_RXQ_CODE_EXIT)) { + rte_pktmbuf_free_seg(pkt); + if (rxq_code == MLX5_RXQ_CODE_DROPPED) { + ++rxq->stats.idropped; + continue; + } + if (rxq_code == MLX5_RXQ_CODE_NOMBUF) { + ++rxq->stats.rx_nombuf; + break; + } + } + rxq_cq_to_mbuf(rxq, pkt, cqe, mcqe); + if (cqe->lro_num_seg > 1) { + mlx5_lro_update_hdr(rte_pktmbuf_mtod(pkt, uint8_t *), + cqe, mcqe, rxq, len); + pkt->ol_flags |= PKT_RX_LRO; + pkt->tso_segsz = len / cqe->lro_num_seg; + } + PKT_LEN(pkt) = len; + PORT(pkt) = rxq->port_id; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Increment bytes counter. */ + rxq->stats.ibytes += PKT_LEN(pkt); +#endif + /* Return packet. */ + *(pkts++) = pkt; + ++i; + } + /* Update the consumer indexes. */ + rxq->consumed_strd = consumed_strd; + rte_io_wmb(); + *rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci); + if (rq_ci != rxq->rq_ci) { + rxq->rq_ci = rq_ci; + rte_io_wmb(); + *rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci); + } +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Increment packets counter. */ + rxq->stats.ipackets += i; +#endif + return i; +} + +/** + * Dummy DPDK callback for RX. + * + * This function is used to temporarily replace the real callback during + * unsafe control operations on the queue, or in case of error. + * + * @param dpdk_rxq + * Generic pointer to RX queue structure. + * @param[out] pkts + * Array to store received packets. + * @param pkts_n + * Maximum number of packets in array. + * + * @return + * Number of packets successfully received (<= pkts_n). + */ +uint16_t +removed_rx_burst(void *dpdk_rxq __rte_unused, + struct rte_mbuf **pkts __rte_unused, + uint16_t pkts_n __rte_unused) +{ + rte_mb(); + return 0; +} + +/* + * Vectorized Rx routines are not compiled in when required vector instructions + * are not supported on a target architecture. + * The following null stubs are needed for linkage when those are not included + * outside of this file (e.g. mlx5_rxtx_vec_sse.c for x86). + */ + +__rte_weak uint16_t +mlx5_rx_burst_vec(void *dpdk_rxq __rte_unused, + struct rte_mbuf **pkts __rte_unused, + uint16_t pkts_n __rte_unused) +{ + return 0; +} + +__rte_weak uint16_t +mlx5_rx_burst_mprq_vec(void *dpdk_rxq __rte_unused, + struct rte_mbuf **pkts __rte_unused, + uint16_t pkts_n __rte_unused) +{ + return 0; +} + +__rte_weak int +mlx5_rxq_check_vec_support(struct mlx5_rxq_data *rxq __rte_unused) +{ + return -ENOTSUP; +} + +__rte_weak int +mlx5_check_vec_rx_support(struct rte_eth_dev *dev __rte_unused) +{ + return -ENOTSUP; +} + diff --git a/drivers/net/mlx5/mlx5_rx.h b/drivers/net/mlx5/mlx5_rx.h index 83b1f38..d5a2de8 100644 --- a/drivers/net/mlx5/mlx5_rx.h +++ b/drivers/net/mlx5/mlx5_rx.h @@ -246,7 +246,7 @@ int mlx5_hrxq_modify(struct rte_eth_dev *dev, uint32_t hxrq_idx, uint64_t hash_fields, const uint16_t *queues, uint32_t queues_n); -/* mlx5_rxtx.c */ +/* mlx5_rx.c */ uint16_t mlx5_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n); void mlx5_rxq_initialize(struct mlx5_rxq_data *rxq); @@ -264,7 +264,7 @@ void mlx5_rxq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, int mlx5_rx_burst_mode_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, struct rte_eth_burst_mode *mode); -/* Vectorized version of mlx5_rxtx.c */ +/* Vectorized version of mlx5_rx.c */ int mlx5_rxq_check_vec_support(struct mlx5_rxq_data *rxq_data); int mlx5_check_vec_rx_support(struct rte_eth_dev *dev); uint16_t mlx5_rx_burst_vec(void *dpdk_rxq, struct rte_mbuf **pkts, diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c index d004e1e..c7f2605 100644 --- a/drivers/net/mlx5/mlx5_rxtx.c +++ b/drivers/net/mlx5/mlx5_rxtx.c @@ -130,37 +130,6 @@ enum mlx5_txcmp_code { static_assert(MLX5_WQE_SIZE == 4 * MLX5_WSEG_SIZE, "invalid WQE size"); -static __rte_always_inline uint32_t -rxq_cq_to_pkt_type(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, - volatile struct mlx5_mini_cqe8 *mcqe); - -static __rte_always_inline int -mlx5_rx_poll_len(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, - uint16_t cqe_cnt, volatile struct mlx5_mini_cqe8 **mcqe); - -static __rte_always_inline uint32_t -rxq_cq_to_ol_flags(volatile struct mlx5_cqe *cqe); - -static __rte_always_inline void -rxq_cq_to_mbuf(struct mlx5_rxq_data *rxq, struct rte_mbuf *pkt, - volatile struct mlx5_cqe *cqe, - volatile struct mlx5_mini_cqe8 *mcqe); - -static int -mlx5_queue_state_modify(struct rte_eth_dev *dev, - struct mlx5_mp_arg_queue_state_modify *sm); - -static inline void -mlx5_lro_update_tcp_hdr(struct rte_tcp_hdr *__rte_restrict tcp, - volatile struct mlx5_cqe *__rte_restrict cqe, - uint32_t phcsum, uint8_t l4_type); - -static inline void -mlx5_lro_update_hdr(uint8_t *__rte_restrict padd, - volatile struct mlx5_cqe *__rte_restrict cqe, - volatile struct mlx5_mini_cqe8 *mcqe, - struct mlx5_rxq_data *rxq, uint32_t len); - uint32_t mlx5_ptype_table[] __rte_cache_aligned = { [0xff] = RTE_PTYPE_ALL_MASK, /* Last entry for errored packet. */ }; @@ -499,220 +468,6 @@ enum mlx5_txcmp_code { return mlx5_cksum_table[idx]; } -/** - * Internal function to compute the number of used descriptors in an RX queue - * - * @param rxq - * The Rx queue. - * - * @return - * The number of used rx descriptor. - */ -static uint32_t -rx_queue_count(struct mlx5_rxq_data *rxq) -{ - struct rxq_zip *zip = &rxq->zip; - volatile struct mlx5_cqe *cqe; - const unsigned int cqe_n = (1 << rxq->cqe_n); - const unsigned int sges_n = (1 << rxq->sges_n); - const unsigned int elts_n = (1 << rxq->elts_n); - const unsigned int strd_n = (1 << rxq->strd_num_n); - const unsigned int cqe_cnt = cqe_n - 1; - unsigned int cq_ci, used; - - /* if we are processing a compressed cqe */ - if (zip->ai) { - used = zip->cqe_cnt - zip->ai; - cq_ci = zip->cq_ci; - } else { - used = 0; - cq_ci = rxq->cq_ci; - } - cqe = &(*rxq->cqes)[cq_ci & cqe_cnt]; - while (check_cqe(cqe, cqe_n, cq_ci) != MLX5_CQE_STATUS_HW_OWN) { - int8_t op_own; - unsigned int n; - - op_own = cqe->op_own; - if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED) - n = rte_be_to_cpu_32(cqe->byte_cnt); - else - n = 1; - cq_ci += n; - used += n; - cqe = &(*rxq->cqes)[cq_ci & cqe_cnt]; - } - used = RTE_MIN(used * sges_n, elts_n * strd_n); - return used; -} - -/** - * DPDK callback to check the status of a rx descriptor. - * - * @param rx_queue - * The Rx queue. - * @param[in] offset - * The index of the descriptor in the ring. - * - * @return - * The status of the tx descriptor. - */ -int -mlx5_rx_descriptor_status(void *rx_queue, uint16_t offset) -{ - struct mlx5_rxq_data *rxq = rx_queue; - struct mlx5_rxq_ctrl *rxq_ctrl = - container_of(rxq, struct mlx5_rxq_ctrl, rxq); - struct rte_eth_dev *dev = ETH_DEV(rxq_ctrl->priv); - - if (dev->rx_pkt_burst == NULL || - dev->rx_pkt_burst == removed_rx_burst) { - rte_errno = ENOTSUP; - return -rte_errno; - } - if (offset >= (1 << rxq->cqe_n)) { - rte_errno = EINVAL; - return -rte_errno; - } - if (offset < rx_queue_count(rxq)) - return RTE_ETH_RX_DESC_DONE; - return RTE_ETH_RX_DESC_AVAIL; -} - -/** - * DPDK callback to get the RX queue information - * - * @param dev - * Pointer to the device structure. - * - * @param rx_queue_id - * Rx queue identificator. - * - * @param qinfo - * Pointer to the RX queue information structure. - * - * @return - * None. - */ - -void -mlx5_rxq_info_get(struct rte_eth_dev *dev, uint16_t rx_queue_id, - struct rte_eth_rxq_info *qinfo) -{ - struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_rxq_data *rxq = (*priv->rxqs)[rx_queue_id]; - struct mlx5_rxq_ctrl *rxq_ctrl = - container_of(rxq, struct mlx5_rxq_ctrl, rxq); - - if (!rxq) - return; - qinfo->mp = mlx5_rxq_mprq_enabled(rxq) ? - rxq->mprq_mp : rxq->mp; - qinfo->conf.rx_thresh.pthresh = 0; - qinfo->conf.rx_thresh.hthresh = 0; - qinfo->conf.rx_thresh.wthresh = 0; - qinfo->conf.rx_free_thresh = rxq->rq_repl_thresh; - qinfo->conf.rx_drop_en = 1; - qinfo->conf.rx_deferred_start = rxq_ctrl ? 0 : 1; - qinfo->conf.offloads = dev->data->dev_conf.rxmode.offloads; - qinfo->scattered_rx = dev->data->scattered_rx; - qinfo->nb_desc = mlx5_rxq_mprq_enabled(rxq) ? - (1 << rxq->elts_n) * (1 << rxq->strd_num_n) : - (1 << rxq->elts_n); -} - -/** - * DPDK callback to get the RX packet burst mode information - * - * @param dev - * Pointer to the device structure. - * - * @param rx_queue_id - * Rx queue identificatior. - * - * @param mode - * Pointer to the burts mode information. - * - * @return - * 0 as success, -EINVAL as failure. - */ - -int -mlx5_rx_burst_mode_get(struct rte_eth_dev *dev, - uint16_t rx_queue_id __rte_unused, - struct rte_eth_burst_mode *mode) -{ - eth_rx_burst_t pkt_burst = dev->rx_pkt_burst; - struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_rxq_data *rxq; - - rxq = (*priv->rxqs)[rx_queue_id]; - if (!rxq) { - rte_errno = EINVAL; - return -rte_errno; - } - if (pkt_burst == mlx5_rx_burst) { - snprintf(mode->info, sizeof(mode->info), "%s", "Scalar"); - } else if (pkt_burst == mlx5_rx_burst_mprq) { - snprintf(mode->info, sizeof(mode->info), "%s", "Multi-Packet RQ"); - } else if (pkt_burst == mlx5_rx_burst_vec) { -#if defined RTE_ARCH_X86_64 - snprintf(mode->info, sizeof(mode->info), "%s", "Vector SSE"); -#elif defined RTE_ARCH_ARM64 - snprintf(mode->info, sizeof(mode->info), "%s", "Vector Neon"); -#elif defined RTE_ARCH_PPC_64 - snprintf(mode->info, sizeof(mode->info), "%s", "Vector AltiVec"); -#else - return -EINVAL; -#endif - } else if (pkt_burst == mlx5_rx_burst_mprq_vec) { -#if defined RTE_ARCH_X86_64 - snprintf(mode->info, sizeof(mode->info), "%s", "MPRQ Vector SSE"); -#elif defined RTE_ARCH_ARM64 - snprintf(mode->info, sizeof(mode->info), "%s", "MPRQ Vector Neon"); -#elif defined RTE_ARCH_PPC_64 - snprintf(mode->info, sizeof(mode->info), "%s", "MPRQ Vector AltiVec"); -#else - return -EINVAL; -#endif - } else { - return -EINVAL; - } - return 0; -} - -/** - * DPDK callback to get the number of used descriptors in a RX queue - * - * @param dev - * Pointer to the device structure. - * - * @param rx_queue_id - * The Rx queue. - * - * @return - * The number of used rx descriptor. - * -EINVAL if the queue is invalid - */ -uint32_t -mlx5_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id) -{ - struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_rxq_data *rxq; - - if (dev->rx_pkt_burst == NULL || - dev->rx_pkt_burst == removed_rx_burst) { - rte_errno = ENOTSUP; - return -rte_errno; - } - rxq = (*priv->rxqs)[rx_queue_id]; - if (!rxq) { - rte_errno = EINVAL; - return -rte_errno; - } - return rx_queue_count(rxq); -} - #define MLX5_SYSTEM_LOG_DIR "/var/log" /** * Dump debug information to log file. @@ -863,101 +618,6 @@ enum mlx5_txcmp_code { } /** - * Translate RX completion flags to packet type. - * - * @param[in] rxq - * Pointer to RX queue structure. - * @param[in] cqe - * Pointer to CQE. - * - * @note: fix mlx5_dev_supported_ptypes_get() if any change here. - * - * @return - * Packet type for struct rte_mbuf. - */ -static inline uint32_t -rxq_cq_to_pkt_type(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, - volatile struct mlx5_mini_cqe8 *mcqe) -{ - uint8_t idx; - uint8_t ptype; - uint8_t pinfo = (cqe->pkt_info & 0x3) << 6; - - /* Get l3/l4 header from mini-CQE in case L3/L4 format*/ - if (mcqe == NULL || - rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_L34H_STRIDX) - ptype = (cqe->hdr_type_etc & 0xfc00) >> 10; - else - ptype = mcqe->hdr_type >> 2; - /* - * The index to the array should have: - * bit[1:0] = l3_hdr_type - * bit[4:2] = l4_hdr_type - * bit[5] = ip_frag - * bit[6] = tunneled - * bit[7] = outer_l3_type - */ - idx = pinfo | ptype; - return mlx5_ptype_table[idx] | rxq->tunnel * !!(idx & (1 << 6)); -} - -/** - * Initialize Rx WQ and indexes. - * - * @param[in] rxq - * Pointer to RX queue structure. - */ -void -mlx5_rxq_initialize(struct mlx5_rxq_data *rxq) -{ - const unsigned int wqe_n = 1 << rxq->elts_n; - unsigned int i; - - for (i = 0; (i != wqe_n); ++i) { - volatile struct mlx5_wqe_data_seg *scat; - uintptr_t addr; - uint32_t byte_count; - - if (mlx5_rxq_mprq_enabled(rxq)) { - struct mlx5_mprq_buf *buf = (*rxq->mprq_bufs)[i]; - - scat = &((volatile struct mlx5_wqe_mprq *) - rxq->wqes)[i].dseg; - addr = (uintptr_t)mlx5_mprq_buf_addr(buf, - 1 << rxq->strd_num_n); - byte_count = (1 << rxq->strd_sz_n) * - (1 << rxq->strd_num_n); - } else { - struct rte_mbuf *buf = (*rxq->elts)[i]; - - scat = &((volatile struct mlx5_wqe_data_seg *) - rxq->wqes)[i]; - addr = rte_pktmbuf_mtod(buf, uintptr_t); - byte_count = DATA_LEN(buf); - } - /* scat->addr must be able to store a pointer. */ - MLX5_ASSERT(sizeof(scat->addr) >= sizeof(uintptr_t)); - *scat = (struct mlx5_wqe_data_seg){ - .addr = rte_cpu_to_be_64(addr), - .byte_count = rte_cpu_to_be_32(byte_count), - .lkey = mlx5_rx_addr2mr(rxq, addr), - }; - } - rxq->consumed_strd = 0; - rxq->decompressed = 0; - rxq->rq_pi = 0; - rxq->zip = (struct rxq_zip){ - .ai = 0, - }; - rxq->elts_ci = mlx5_rxq_mprq_enabled(rxq) ? - (wqe_n >> rxq->sges_n) * (1 << rxq->strd_num_n) : 0; - /* Update doorbell counter. */ - rxq->rq_ci = wqe_n >> rxq->sges_n; - rte_io_wmb(); - *rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci); -} - -/** * Modify a Verbs/DevX queue state. * This must be called from the primary process. * @@ -1013,7 +673,7 @@ enum mlx5_txcmp_code { * @return * 0 in case of success else non-zero value. */ -static int +int mlx5_queue_state_modify(struct rte_eth_dev *dev, struct mlx5_mp_arg_queue_state_modify *sm) { @@ -1034,783 +694,6 @@ enum mlx5_txcmp_code { } /** - * Handle a Rx error. - * The function inserts the RQ state to reset when the first error CQE is - * shown, then drains the CQ by the caller function loop. When the CQ is empty, - * it moves the RQ state to ready and initializes the RQ. - * Next CQE identification and error counting are in the caller responsibility. - * - * @param[in] rxq - * Pointer to RX queue structure. - * @param[in] vec - * 1 when called from vectorized Rx burst, need to prepare mbufs for the RQ. - * 0 when called from non-vectorized Rx burst. - * - * @return - * -1 in case of recovery error, otherwise the CQE status. - */ -int -mlx5_rx_err_handle(struct mlx5_rxq_data *rxq, uint8_t vec) -{ - const uint16_t cqe_n = 1 << rxq->cqe_n; - const uint16_t cqe_mask = cqe_n - 1; - const uint16_t wqe_n = 1 << rxq->elts_n; - const uint16_t strd_n = 1 << rxq->strd_num_n; - struct mlx5_rxq_ctrl *rxq_ctrl = - container_of(rxq, struct mlx5_rxq_ctrl, rxq); - union { - volatile struct mlx5_cqe *cqe; - volatile struct mlx5_err_cqe *err_cqe; - } u = { - .cqe = &(*rxq->cqes)[rxq->cq_ci & cqe_mask], - }; - struct mlx5_mp_arg_queue_state_modify sm; - int ret; - - switch (rxq->err_state) { - case MLX5_RXQ_ERR_STATE_NO_ERROR: - rxq->err_state = MLX5_RXQ_ERR_STATE_NEED_RESET; - /* Fall-through */ - case MLX5_RXQ_ERR_STATE_NEED_RESET: - sm.is_wq = 1; - sm.queue_id = rxq->idx; - sm.state = IBV_WQS_RESET; - if (mlx5_queue_state_modify(ETH_DEV(rxq_ctrl->priv), &sm)) - return -1; - if (rxq_ctrl->dump_file_n < - rxq_ctrl->priv->config.max_dump_files_num) { - MKSTR(err_str, "Unexpected CQE error syndrome " - "0x%02x CQN = %u RQN = %u wqe_counter = %u" - " rq_ci = %u cq_ci = %u", u.err_cqe->syndrome, - rxq->cqn, rxq_ctrl->wqn, - rte_be_to_cpu_16(u.err_cqe->wqe_counter), - rxq->rq_ci << rxq->sges_n, rxq->cq_ci); - MKSTR(name, "dpdk_mlx5_port_%u_rxq_%u_%u", - rxq->port_id, rxq->idx, (uint32_t)rte_rdtsc()); - mlx5_dump_debug_information(name, NULL, err_str, 0); - mlx5_dump_debug_information(name, "MLX5 Error CQ:", - (const void *)((uintptr_t) - rxq->cqes), - sizeof(*u.cqe) * cqe_n); - mlx5_dump_debug_information(name, "MLX5 Error RQ:", - (const void *)((uintptr_t) - rxq->wqes), - 16 * wqe_n); - rxq_ctrl->dump_file_n++; - } - rxq->err_state = MLX5_RXQ_ERR_STATE_NEED_READY; - /* Fall-through */ - case MLX5_RXQ_ERR_STATE_NEED_READY: - ret = check_cqe(u.cqe, cqe_n, rxq->cq_ci); - if (ret == MLX5_CQE_STATUS_HW_OWN) { - rte_io_wmb(); - *rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci); - rte_io_wmb(); - /* - * The RQ consumer index must be zeroed while moving - * from RESET state to RDY state. - */ - *rxq->rq_db = rte_cpu_to_be_32(0); - rte_io_wmb(); - sm.is_wq = 1; - sm.queue_id = rxq->idx; - sm.state = IBV_WQS_RDY; - if (mlx5_queue_state_modify(ETH_DEV(rxq_ctrl->priv), - &sm)) - return -1; - if (vec) { - const uint32_t elts_n = - mlx5_rxq_mprq_enabled(rxq) ? - wqe_n * strd_n : wqe_n; - const uint32_t e_mask = elts_n - 1; - uint32_t elts_ci = - mlx5_rxq_mprq_enabled(rxq) ? - rxq->elts_ci : rxq->rq_ci; - uint32_t elt_idx; - struct rte_mbuf **elt; - int i; - unsigned int n = elts_n - (elts_ci - - rxq->rq_pi); - - for (i = 0; i < (int)n; ++i) { - elt_idx = (elts_ci + i) & e_mask; - elt = &(*rxq->elts)[elt_idx]; - *elt = rte_mbuf_raw_alloc(rxq->mp); - if (!*elt) { - for (i--; i >= 0; --i) { - elt_idx = (elts_ci + - i) & elts_n; - elt = &(*rxq->elts) - [elt_idx]; - rte_pktmbuf_free_seg - (*elt); - } - return -1; - } - } - for (i = 0; i < (int)elts_n; ++i) { - elt = &(*rxq->elts)[i]; - DATA_LEN(*elt) = - (uint16_t)((*elt)->buf_len - - rte_pktmbuf_headroom(*elt)); - } - /* Padding with a fake mbuf for vec Rx. */ - for (i = 0; i < MLX5_VPMD_DESCS_PER_LOOP; ++i) - (*rxq->elts)[elts_n + i] = - &rxq->fake_mbuf; - } - mlx5_rxq_initialize(rxq); - rxq->err_state = MLX5_RXQ_ERR_STATE_NO_ERROR; - } - return ret; - default: - return -1; - } -} - -/** - * Get size of the next packet for a given CQE. For compressed CQEs, the - * consumer index is updated only once all packets of the current one have - * been processed. - * - * @param rxq - * Pointer to RX queue. - * @param cqe - * CQE to process. - * @param[out] mcqe - * Store pointer to mini-CQE if compressed. Otherwise, the pointer is not - * written. - * - * @return - * 0 in case of empty CQE, otherwise the packet size in bytes. - */ -static inline int -mlx5_rx_poll_len(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cqe, - uint16_t cqe_cnt, volatile struct mlx5_mini_cqe8 **mcqe) -{ - struct rxq_zip *zip = &rxq->zip; - uint16_t cqe_n = cqe_cnt + 1; - int len; - uint16_t idx, end; - - do { - len = 0; - /* Process compressed data in the CQE and mini arrays. */ - if (zip->ai) { - volatile struct mlx5_mini_cqe8 (*mc)[8] = - (volatile struct mlx5_mini_cqe8 (*)[8]) - (uintptr_t)(&(*rxq->cqes)[zip->ca & - cqe_cnt].pkt_info); - len = rte_be_to_cpu_32((*mc)[zip->ai & 7].byte_cnt & - rxq->byte_mask); - *mcqe = &(*mc)[zip->ai & 7]; - if ((++zip->ai & 7) == 0) { - /* Invalidate consumed CQEs */ - idx = zip->ca; - end = zip->na; - while (idx != end) { - (*rxq->cqes)[idx & cqe_cnt].op_own = - MLX5_CQE_INVALIDATE; - ++idx; - } - /* - * Increment consumer index to skip the number - * of CQEs consumed. Hardware leaves holes in - * the CQ ring for software use. - */ - zip->ca = zip->na; - zip->na += 8; - } - if (unlikely(rxq->zip.ai == rxq->zip.cqe_cnt)) { - /* Invalidate the rest */ - idx = zip->ca; - end = zip->cq_ci; - - while (idx != end) { - (*rxq->cqes)[idx & cqe_cnt].op_own = - MLX5_CQE_INVALIDATE; - ++idx; - } - rxq->cq_ci = zip->cq_ci; - zip->ai = 0; - } - /* - * No compressed data, get next CQE and verify if it is - * compressed. - */ - } else { - int ret; - int8_t op_own; - uint32_t cq_ci; - - ret = check_cqe(cqe, cqe_n, rxq->cq_ci); - if (unlikely(ret != MLX5_CQE_STATUS_SW_OWN)) { - if (unlikely(ret == MLX5_CQE_STATUS_ERR || - rxq->err_state)) { - ret = mlx5_rx_err_handle(rxq, 0); - if (ret == MLX5_CQE_STATUS_HW_OWN || - ret == -1) - return 0; - } else { - return 0; - } - } - /* - * Introduce the local variable to have queue cq_ci - * index in queue structure always consistent with - * actual CQE boundary (not pointing to the middle - * of compressed CQE session). - */ - cq_ci = rxq->cq_ci + 1; - op_own = cqe->op_own; - if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED) { - volatile struct mlx5_mini_cqe8 (*mc)[8] = - (volatile struct mlx5_mini_cqe8 (*)[8]) - (uintptr_t)(&(*rxq->cqes) - [cq_ci & cqe_cnt].pkt_info); - - /* Fix endianness. */ - zip->cqe_cnt = rte_be_to_cpu_32(cqe->byte_cnt); - /* - * Current mini array position is the one - * returned by check_cqe64(). - * - * If completion comprises several mini arrays, - * as a special case the second one is located - * 7 CQEs after the initial CQE instead of 8 - * for subsequent ones. - */ - zip->ca = cq_ci; - zip->na = zip->ca + 7; - /* Compute the next non compressed CQE. */ - zip->cq_ci = rxq->cq_ci + zip->cqe_cnt; - /* Get packet size to return. */ - len = rte_be_to_cpu_32((*mc)[0].byte_cnt & - rxq->byte_mask); - *mcqe = &(*mc)[0]; - zip->ai = 1; - /* Prefetch all to be invalidated */ - idx = zip->ca; - end = zip->cq_ci; - while (idx != end) { - rte_prefetch0(&(*rxq->cqes)[(idx) & - cqe_cnt]); - ++idx; - } - } else { - rxq->cq_ci = cq_ci; - len = rte_be_to_cpu_32(cqe->byte_cnt); - } - } - if (unlikely(rxq->err_state)) { - cqe = &(*rxq->cqes)[rxq->cq_ci & cqe_cnt]; - ++rxq->stats.idropped; - } else { - return len; - } - } while (1); -} - -/** - * Translate RX completion flags to offload flags. - * - * @param[in] cqe - * Pointer to CQE. - * - * @return - * Offload flags (ol_flags) for struct rte_mbuf. - */ -static inline uint32_t -rxq_cq_to_ol_flags(volatile struct mlx5_cqe *cqe) -{ - uint32_t ol_flags = 0; - uint16_t flags = rte_be_to_cpu_16(cqe->hdr_type_etc); - - ol_flags = - TRANSPOSE(flags, - MLX5_CQE_RX_L3_HDR_VALID, - PKT_RX_IP_CKSUM_GOOD) | - TRANSPOSE(flags, - MLX5_CQE_RX_L4_HDR_VALID, - PKT_RX_L4_CKSUM_GOOD); - return ol_flags; -} - -/** - * Fill in mbuf fields from RX completion flags. - * Note that pkt->ol_flags should be initialized outside of this function. - * - * @param rxq - * Pointer to RX queue. - * @param pkt - * mbuf to fill. - * @param cqe - * CQE to process. - * @param rss_hash_res - * Packet RSS Hash result. - */ -static inline void -rxq_cq_to_mbuf(struct mlx5_rxq_data *rxq, struct rte_mbuf *pkt, - volatile struct mlx5_cqe *cqe, - volatile struct mlx5_mini_cqe8 *mcqe) -{ - /* Update packet information. */ - pkt->packet_type = rxq_cq_to_pkt_type(rxq, cqe, mcqe); - - if (rxq->rss_hash) { - uint32_t rss_hash_res = 0; - - /* If compressed, take hash result from mini-CQE. */ - if (mcqe == NULL || - rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_HASH) - rss_hash_res = rte_be_to_cpu_32(cqe->rx_hash_res); - else - rss_hash_res = rte_be_to_cpu_32(mcqe->rx_hash_result); - if (rss_hash_res) { - pkt->hash.rss = rss_hash_res; - pkt->ol_flags |= PKT_RX_RSS_HASH; - } - } - if (rxq->mark) { - uint32_t mark = 0; - - /* If compressed, take flow tag from mini-CQE. */ - if (mcqe == NULL || - rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_FTAG_STRIDX) - mark = cqe->sop_drop_qpn; - else - mark = ((mcqe->byte_cnt_flow & 0xff) << 8) | - (mcqe->flow_tag_high << 16); - if (MLX5_FLOW_MARK_IS_VALID(mark)) { - pkt->ol_flags |= PKT_RX_FDIR; - if (mark != RTE_BE32(MLX5_FLOW_MARK_DEFAULT)) { - pkt->ol_flags |= PKT_RX_FDIR_ID; - pkt->hash.fdir.hi = mlx5_flow_mark_get(mark); - } - } - } - if (rxq->dynf_meta) { - uint32_t meta = cqe->flow_table_metadata & - rxq->flow_meta_port_mask; - - if (meta) { - pkt->ol_flags |= rxq->flow_meta_mask; - *RTE_MBUF_DYNFIELD(pkt, rxq->flow_meta_offset, - uint32_t *) = meta; - } - } - if (rxq->csum) - pkt->ol_flags |= rxq_cq_to_ol_flags(cqe); - if (rxq->vlan_strip) { - bool vlan_strip; - - if (mcqe == NULL || - rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_L34H_STRIDX) - vlan_strip = cqe->hdr_type_etc & - RTE_BE16(MLX5_CQE_VLAN_STRIPPED); - else - vlan_strip = mcqe->hdr_type & - RTE_BE16(MLX5_CQE_VLAN_STRIPPED); - if (vlan_strip) { - pkt->ol_flags |= PKT_RX_VLAN | PKT_RX_VLAN_STRIPPED; - pkt->vlan_tci = rte_be_to_cpu_16(cqe->vlan_info); - } - } - if (rxq->hw_timestamp) { - uint64_t ts = rte_be_to_cpu_64(cqe->timestamp); - - if (rxq->rt_timestamp) - ts = mlx5_txpp_convert_rx_ts(rxq->sh, ts); - mlx5_timestamp_set(pkt, rxq->timestamp_offset, ts); - pkt->ol_flags |= rxq->timestamp_rx_flag; - } -} - -/** - * DPDK callback for RX. - * - * @param dpdk_rxq - * Generic pointer to RX queue structure. - * @param[out] pkts - * Array to store received packets. - * @param pkts_n - * Maximum number of packets in array. - * - * @return - * Number of packets successfully received (<= pkts_n). - */ -uint16_t -mlx5_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n) -{ - struct mlx5_rxq_data *rxq = dpdk_rxq; - const unsigned int wqe_cnt = (1 << rxq->elts_n) - 1; - const unsigned int cqe_cnt = (1 << rxq->cqe_n) - 1; - const unsigned int sges_n = rxq->sges_n; - struct rte_mbuf *pkt = NULL; - struct rte_mbuf *seg = NULL; - volatile struct mlx5_cqe *cqe = - &(*rxq->cqes)[rxq->cq_ci & cqe_cnt]; - unsigned int i = 0; - unsigned int rq_ci = rxq->rq_ci << sges_n; - int len = 0; /* keep its value across iterations. */ - - while (pkts_n) { - unsigned int idx = rq_ci & wqe_cnt; - volatile struct mlx5_wqe_data_seg *wqe = - &((volatile struct mlx5_wqe_data_seg *)rxq->wqes)[idx]; - struct rte_mbuf *rep = (*rxq->elts)[idx]; - volatile struct mlx5_mini_cqe8 *mcqe = NULL; - - if (pkt) - NEXT(seg) = rep; - seg = rep; - rte_prefetch0(seg); - rte_prefetch0(cqe); - rte_prefetch0(wqe); - /* Allocate the buf from the same pool. */ - rep = rte_mbuf_raw_alloc(seg->pool); - if (unlikely(rep == NULL)) { - ++rxq->stats.rx_nombuf; - if (!pkt) { - /* - * no buffers before we even started, - * bail out silently. - */ - break; - } - while (pkt != seg) { - MLX5_ASSERT(pkt != (*rxq->elts)[idx]); - rep = NEXT(pkt); - NEXT(pkt) = NULL; - NB_SEGS(pkt) = 1; - rte_mbuf_raw_free(pkt); - pkt = rep; - } - rq_ci >>= sges_n; - ++rq_ci; - rq_ci <<= sges_n; - break; - } - if (!pkt) { - cqe = &(*rxq->cqes)[rxq->cq_ci & cqe_cnt]; - len = mlx5_rx_poll_len(rxq, cqe, cqe_cnt, &mcqe); - if (!len) { - rte_mbuf_raw_free(rep); - break; - } - pkt = seg; - MLX5_ASSERT(len >= (rxq->crc_present << 2)); - pkt->ol_flags &= EXT_ATTACHED_MBUF; - rxq_cq_to_mbuf(rxq, pkt, cqe, mcqe); - if (rxq->crc_present) - len -= RTE_ETHER_CRC_LEN; - PKT_LEN(pkt) = len; - if (cqe->lro_num_seg > 1) { - mlx5_lro_update_hdr - (rte_pktmbuf_mtod(pkt, uint8_t *), cqe, - mcqe, rxq, len); - pkt->ol_flags |= PKT_RX_LRO; - pkt->tso_segsz = len / cqe->lro_num_seg; - } - } - DATA_LEN(rep) = DATA_LEN(seg); - PKT_LEN(rep) = PKT_LEN(seg); - SET_DATA_OFF(rep, DATA_OFF(seg)); - PORT(rep) = PORT(seg); - (*rxq->elts)[idx] = rep; - /* - * Fill NIC descriptor with the new buffer. The lkey and size - * of the buffers are already known, only the buffer address - * changes. - */ - wqe->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t)); - /* If there's only one MR, no need to replace LKey in WQE. */ - if (unlikely(mlx5_mr_btree_len(&rxq->mr_ctrl.cache_bh) > 1)) - wqe->lkey = mlx5_rx_mb2mr(rxq, rep); - if (len > DATA_LEN(seg)) { - len -= DATA_LEN(seg); - ++NB_SEGS(pkt); - ++rq_ci; - continue; - } - DATA_LEN(seg) = len; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Increment bytes counter. */ - rxq->stats.ibytes += PKT_LEN(pkt); -#endif - /* Return packet. */ - *(pkts++) = pkt; - pkt = NULL; - --pkts_n; - ++i; - /* Align consumer index to the next stride. */ - rq_ci >>= sges_n; - ++rq_ci; - rq_ci <<= sges_n; - } - if (unlikely((i == 0) && ((rq_ci >> sges_n) == rxq->rq_ci))) - return 0; - /* Update the consumer index. */ - rxq->rq_ci = rq_ci >> sges_n; - rte_io_wmb(); - *rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci); - rte_io_wmb(); - *rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci); -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Increment packets counter. */ - rxq->stats.ipackets += i; -#endif - return i; -} - -/** - * Update LRO packet TCP header. - * The HW LRO feature doesn't update the TCP header after coalescing the - * TCP segments but supplies information in CQE to fill it by SW. - * - * @param tcp - * Pointer to the TCP header. - * @param cqe - * Pointer to the completion entry.. - * @param phcsum - * The L3 pseudo-header checksum. - */ -static inline void -mlx5_lro_update_tcp_hdr(struct rte_tcp_hdr *__rte_restrict tcp, - volatile struct mlx5_cqe *__rte_restrict cqe, - uint32_t phcsum, uint8_t l4_type) -{ - /* - * The HW calculates only the TCP payload checksum, need to complete - * the TCP header checksum and the L3 pseudo-header checksum. - */ - uint32_t csum = phcsum + cqe->csum; - - if (l4_type == MLX5_L4_HDR_TYPE_TCP_EMPTY_ACK || - l4_type == MLX5_L4_HDR_TYPE_TCP_WITH_ACL) { - tcp->tcp_flags |= RTE_TCP_ACK_FLAG; - tcp->recv_ack = cqe->lro_ack_seq_num; - tcp->rx_win = cqe->lro_tcp_win; - } - if (cqe->lro_tcppsh_abort_dupack & MLX5_CQE_LRO_PUSH_MASK) - tcp->tcp_flags |= RTE_TCP_PSH_FLAG; - tcp->cksum = 0; - csum += rte_raw_cksum(tcp, (tcp->data_off >> 4) * 4); - csum = ((csum & 0xffff0000) >> 16) + (csum & 0xffff); - csum = (~csum) & 0xffff; - if (csum == 0) - csum = 0xffff; - tcp->cksum = csum; -} - -/** - * Update LRO packet headers. - * The HW LRO feature doesn't update the L3/TCP headers after coalescing the - * TCP segments but supply information in CQE to fill it by SW. - * - * @param padd - * The packet address. - * @param cqe - * Pointer to the completion entry.. - * @param len - * The packet length. - */ -static inline void -mlx5_lro_update_hdr(uint8_t *__rte_restrict padd, - volatile struct mlx5_cqe *__rte_restrict cqe, - volatile struct mlx5_mini_cqe8 *mcqe, - struct mlx5_rxq_data *rxq, uint32_t len) -{ - union { - struct rte_ether_hdr *eth; - struct rte_vlan_hdr *vlan; - struct rte_ipv4_hdr *ipv4; - struct rte_ipv6_hdr *ipv6; - struct rte_tcp_hdr *tcp; - uint8_t *hdr; - } h = { - .hdr = padd, - }; - uint16_t proto = h.eth->ether_type; - uint32_t phcsum; - uint8_t l4_type; - - h.eth++; - while (proto == RTE_BE16(RTE_ETHER_TYPE_VLAN) || - proto == RTE_BE16(RTE_ETHER_TYPE_QINQ)) { - proto = h.vlan->eth_proto; - h.vlan++; - } - if (proto == RTE_BE16(RTE_ETHER_TYPE_IPV4)) { - h.ipv4->time_to_live = cqe->lro_min_ttl; - h.ipv4->total_length = rte_cpu_to_be_16(len - (h.hdr - padd)); - h.ipv4->hdr_checksum = 0; - h.ipv4->hdr_checksum = rte_ipv4_cksum(h.ipv4); - phcsum = rte_ipv4_phdr_cksum(h.ipv4, 0); - h.ipv4++; - } else { - h.ipv6->hop_limits = cqe->lro_min_ttl; - h.ipv6->payload_len = rte_cpu_to_be_16(len - (h.hdr - padd) - - sizeof(*h.ipv6)); - phcsum = rte_ipv6_phdr_cksum(h.ipv6, 0); - h.ipv6++; - } - if (mcqe == NULL || - rxq->mcqe_format != MLX5_CQE_RESP_FORMAT_L34H_STRIDX) - l4_type = (rte_be_to_cpu_16(cqe->hdr_type_etc) & - MLX5_CQE_L4_TYPE_MASK) >> MLX5_CQE_L4_TYPE_SHIFT; - else - l4_type = (rte_be_to_cpu_16(mcqe->hdr_type) & - MLX5_CQE_L4_TYPE_MASK) >> MLX5_CQE_L4_TYPE_SHIFT; - mlx5_lro_update_tcp_hdr(h.tcp, cqe, phcsum, l4_type); -} - -void -mlx5_mprq_buf_free_cb(void *addr __rte_unused, void *opaque) -{ - struct mlx5_mprq_buf *buf = opaque; - - if (__atomic_load_n(&buf->refcnt, __ATOMIC_RELAXED) == 1) { - rte_mempool_put(buf->mp, buf); - } else if (unlikely(__atomic_sub_fetch(&buf->refcnt, 1, - __ATOMIC_RELAXED) == 0)) { - __atomic_store_n(&buf->refcnt, 1, __ATOMIC_RELAXED); - rte_mempool_put(buf->mp, buf); - } -} - -void -mlx5_mprq_buf_free(struct mlx5_mprq_buf *buf) -{ - mlx5_mprq_buf_free_cb(NULL, buf); -} - -/** - * DPDK callback for RX with Multi-Packet RQ support. - * - * @param dpdk_rxq - * Generic pointer to RX queue structure. - * @param[out] pkts - * Array to store received packets. - * @param pkts_n - * Maximum number of packets in array. - * - * @return - * Number of packets successfully received (<= pkts_n). - */ -uint16_t -mlx5_rx_burst_mprq(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n) -{ - struct mlx5_rxq_data *rxq = dpdk_rxq; - const uint32_t strd_n = 1 << rxq->strd_num_n; - const uint32_t strd_sz = 1 << rxq->strd_sz_n; - const uint32_t cq_mask = (1 << rxq->cqe_n) - 1; - const uint32_t wq_mask = (1 << rxq->elts_n) - 1; - volatile struct mlx5_cqe *cqe = &(*rxq->cqes)[rxq->cq_ci & cq_mask]; - unsigned int i = 0; - uint32_t rq_ci = rxq->rq_ci; - uint16_t consumed_strd = rxq->consumed_strd; - struct mlx5_mprq_buf *buf = (*rxq->mprq_bufs)[rq_ci & wq_mask]; - - while (i < pkts_n) { - struct rte_mbuf *pkt; - int ret; - uint32_t len; - uint16_t strd_cnt; - uint16_t strd_idx; - uint32_t byte_cnt; - volatile struct mlx5_mini_cqe8 *mcqe = NULL; - enum mlx5_rqx_code rxq_code; - - if (consumed_strd == strd_n) { - /* Replace WQE if the buffer is still in use. */ - mprq_buf_replace(rxq, rq_ci & wq_mask); - /* Advance to the next WQE. */ - consumed_strd = 0; - ++rq_ci; - buf = (*rxq->mprq_bufs)[rq_ci & wq_mask]; - } - cqe = &(*rxq->cqes)[rxq->cq_ci & cq_mask]; - ret = mlx5_rx_poll_len(rxq, cqe, cq_mask, &mcqe); - if (!ret) - break; - byte_cnt = ret; - len = (byte_cnt & MLX5_MPRQ_LEN_MASK) >> MLX5_MPRQ_LEN_SHIFT; - MLX5_ASSERT((int)len >= (rxq->crc_present << 2)); - if (rxq->crc_present) - len -= RTE_ETHER_CRC_LEN; - if (mcqe && - rxq->mcqe_format == MLX5_CQE_RESP_FORMAT_FTAG_STRIDX) - strd_cnt = (len / strd_sz) + !!(len % strd_sz); - else - strd_cnt = (byte_cnt & MLX5_MPRQ_STRIDE_NUM_MASK) >> - MLX5_MPRQ_STRIDE_NUM_SHIFT; - MLX5_ASSERT(strd_cnt); - consumed_strd += strd_cnt; - if (byte_cnt & MLX5_MPRQ_FILLER_MASK) - continue; - strd_idx = rte_be_to_cpu_16(mcqe == NULL ? - cqe->wqe_counter : - mcqe->stride_idx); - MLX5_ASSERT(strd_idx < strd_n); - MLX5_ASSERT(!((rte_be_to_cpu_16(cqe->wqe_id) ^ rq_ci) & - wq_mask)); - pkt = rte_pktmbuf_alloc(rxq->mp); - if (unlikely(pkt == NULL)) { - ++rxq->stats.rx_nombuf; - break; - } - len = (byte_cnt & MLX5_MPRQ_LEN_MASK) >> MLX5_MPRQ_LEN_SHIFT; - MLX5_ASSERT((int)len >= (rxq->crc_present << 2)); - if (rxq->crc_present) - len -= RTE_ETHER_CRC_LEN; - rxq_code = mprq_buf_to_pkt(rxq, pkt, len, buf, - strd_idx, strd_cnt); - if (unlikely(rxq_code != MLX5_RXQ_CODE_EXIT)) { - rte_pktmbuf_free_seg(pkt); - if (rxq_code == MLX5_RXQ_CODE_DROPPED) { - ++rxq->stats.idropped; - continue; - } - if (rxq_code == MLX5_RXQ_CODE_NOMBUF) { - ++rxq->stats.rx_nombuf; - break; - } - } - rxq_cq_to_mbuf(rxq, pkt, cqe, mcqe); - if (cqe->lro_num_seg > 1) { - mlx5_lro_update_hdr(rte_pktmbuf_mtod(pkt, uint8_t *), - cqe, mcqe, rxq, len); - pkt->ol_flags |= PKT_RX_LRO; - pkt->tso_segsz = len / cqe->lro_num_seg; - } - PKT_LEN(pkt) = len; - PORT(pkt) = rxq->port_id; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Increment bytes counter. */ - rxq->stats.ibytes += PKT_LEN(pkt); -#endif - /* Return packet. */ - *(pkts++) = pkt; - ++i; - } - /* Update the consumer indexes. */ - rxq->consumed_strd = consumed_strd; - rte_io_wmb(); - *rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci); - if (rq_ci != rxq->rq_ci) { - rxq->rq_ci = rq_ci; - rte_io_wmb(); - *rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci); - } -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Increment packets counter. */ - rxq->stats.ipackets += i; -#endif - return i; -} - -/** * Dummy DPDK callback for TX. * * This function is used to temporarily replace the real callback during @@ -1836,66 +719,6 @@ enum mlx5_txcmp_code { } /** - * Dummy DPDK callback for RX. - * - * This function is used to temporarily replace the real callback during - * unsafe control operations on the queue, or in case of error. - * - * @param dpdk_rxq - * Generic pointer to RX queue structure. - * @param[out] pkts - * Array to store received packets. - * @param pkts_n - * Maximum number of packets in array. - * - * @return - * Number of packets successfully received (<= pkts_n). - */ -uint16_t -removed_rx_burst(void *dpdk_txq __rte_unused, - struct rte_mbuf **pkts __rte_unused, - uint16_t pkts_n __rte_unused) -{ - rte_mb(); - return 0; -} - -/* - * Vectorized Rx/Tx routines are not compiled in when required vector - * instructions are not supported on a target architecture. The following null - * stubs are needed for linkage when those are not included outside of this file - * (e.g. mlx5_rxtx_vec_sse.c for x86). - */ - -__rte_weak uint16_t -mlx5_rx_burst_vec(void *dpdk_txq __rte_unused, - struct rte_mbuf **pkts __rte_unused, - uint16_t pkts_n __rte_unused) -{ - return 0; -} - -__rte_weak uint16_t -mlx5_rx_burst_mprq_vec(void *dpdk_txq __rte_unused, - struct rte_mbuf **pkts __rte_unused, - uint16_t pkts_n __rte_unused) -{ - return 0; -} - -__rte_weak int -mlx5_rxq_check_vec_support(struct mlx5_rxq_data *rxq __rte_unused) -{ - return -ENOTSUP; -} - -__rte_weak int -mlx5_check_vec_rx_support(struct rte_eth_dev *dev __rte_unused) -{ - return -ENOTSUP; -} - -/** * Free the mbufs from the linear array of pointers. * * @param txq diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h index d443db4..f1ebc99 100644 --- a/drivers/net/mlx5/mlx5_rxtx.h +++ b/drivers/net/mlx5/mlx5_rxtx.h @@ -196,6 +196,8 @@ void mlx5_dump_debug_information(const char *path, const char *title, const void *buf, unsigned int len); int mlx5_queue_state_modify_primary(struct rte_eth_dev *dev, const struct mlx5_mp_arg_queue_state_modify *sm); +int mlx5_queue_state_modify(struct rte_eth_dev *dev, + struct mlx5_mp_arg_queue_state_modify *sm); void mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, struct rte_eth_txq_info *qinfo); int mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, From patchwork Mon Apr 5 14:00:53 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Baum X-Patchwork-Id: 90559 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 71B46A0A02; Mon, 5 Apr 2021 16:01:44 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id AE3EC140EE4; Mon, 5 Apr 2021 16:01:37 +0200 (CEST) Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by mails.dpdk.org (Postfix) with ESMTP id A10DE140EA9 for ; Mon, 5 Apr 2021 16:01:35 +0200 (CEST) Received: from Internal Mail-Server by MTLPINE1 (envelope-from michaelba@nvidia.com) with SMTP; 5 Apr 2021 17:01:32 +0300 Received: from nvidia.com (pegasus07.mtr.labs.mlnx [10.210.16.112]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 135E1VNZ031476; Mon, 5 Apr 2021 17:01:32 +0300 From: Michael Baum To: dev@dpdk.org Cc: Matan Azrad , Raslan Darawsheh , Viacheslav Ovsiienko Date: Mon, 5 Apr 2021 14:00:53 +0000 Message-Id: <1617631256-3018-4-git-send-email-michaelba@nvidia.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> References: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> Subject: [dpdk-dev] [PATCH 3/6] net/mlx5: separate Tx function declarations to another file X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" This patch separates Tx function declarations to different header file in preparation for removing their implementation from the source file and as an optional preparation for Tx cleanup. Signed-off-by: Michael Baum Acked-by: Viacheslav Ovsiienko --- drivers/net/mlx5/linux/mlx5_mp_os.c | 1 + drivers/net/mlx5/linux/mlx5_os.c | 1 + drivers/net/mlx5/linux/mlx5_verbs.c | 2 +- drivers/net/mlx5/mlx5.c | 1 + drivers/net/mlx5/mlx5_devx.c | 2 +- drivers/net/mlx5/mlx5_ethdev.c | 1 + drivers/net/mlx5/mlx5_flow.c | 2 +- drivers/net/mlx5/mlx5_flow_dv.c | 2 +- drivers/net/mlx5/mlx5_flow_verbs.c | 1 - drivers/net/mlx5/mlx5_mr.c | 1 + drivers/net/mlx5/mlx5_rxmode.c | 1 - drivers/net/mlx5/mlx5_rxq.c | 2 +- drivers/net/mlx5/mlx5_rxtx.c | 1 + drivers/net/mlx5/mlx5_rxtx.h | 344 --------------------------------- drivers/net/mlx5/mlx5_stats.c | 2 +- drivers/net/mlx5/mlx5_trigger.c | 2 +- drivers/net/mlx5/mlx5_tx.h | 371 ++++++++++++++++++++++++++++++++++++ drivers/net/mlx5/mlx5_txpp.c | 2 +- drivers/net/mlx5/mlx5_txq.c | 2 +- drivers/net/mlx5/windows/mlx5_os.c | 1 + 20 files changed, 387 insertions(+), 355 deletions(-) create mode 100644 drivers/net/mlx5/mlx5_tx.h diff --git a/drivers/net/mlx5/linux/mlx5_mp_os.c b/drivers/net/mlx5/linux/mlx5_mp_os.c index 63fa278..ca529b6 100644 --- a/drivers/net/mlx5/linux/mlx5_mp_os.c +++ b/drivers/net/mlx5/linux/mlx5_mp_os.c @@ -17,6 +17,7 @@ #include "mlx5.h" #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_utils.h" int diff --git a/drivers/net/mlx5/linux/mlx5_os.c b/drivers/net/mlx5/linux/mlx5_os.c index 97a28ec..026423b 100644 --- a/drivers/net/mlx5/linux/mlx5_os.c +++ b/drivers/net/mlx5/linux/mlx5_os.c @@ -41,6 +41,7 @@ #include "mlx5_utils.h" #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" #include "mlx5_flow.h" diff --git a/drivers/net/mlx5/linux/mlx5_verbs.c b/drivers/net/mlx5/linux/mlx5_verbs.c index 73096af..0b0759f 100644 --- a/drivers/net/mlx5/linux/mlx5_verbs.c +++ b/drivers/net/mlx5/linux/mlx5_verbs.c @@ -20,9 +20,9 @@ #include #include #include -#include #include #include +#include #include #include diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c index 6f77bc2..02cc2c7 100644 --- a/drivers/net/mlx5/mlx5.c +++ b/drivers/net/mlx5/mlx5.c @@ -36,6 +36,7 @@ #include "mlx5_utils.h" #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" #include "mlx5_flow.h" diff --git a/drivers/net/mlx5/mlx5_devx.c b/drivers/net/mlx5/mlx5_devx.c index 76935f6..76d31f5 100644 --- a/drivers/net/mlx5/mlx5_devx.c +++ b/drivers/net/mlx5/mlx5_devx.c @@ -20,7 +20,7 @@ #include "mlx5.h" #include "mlx5_common_os.h" -#include "mlx5_rxtx.h" +#include "mlx5_tx.h" #include "mlx5_rx.h" #include "mlx5_utils.h" #include "mlx5_devx.h" diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c index 708e3a3..90baee5 100644 --- a/drivers/net/mlx5/mlx5_ethdev.c +++ b/drivers/net/mlx5/mlx5_ethdev.c @@ -24,6 +24,7 @@ #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_autoconf.h" /** diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c index b3877a1..4dea006 100644 --- a/drivers/net/mlx5/mlx5_flow.c +++ b/drivers/net/mlx5/mlx5_flow.c @@ -29,8 +29,8 @@ #include "mlx5.h" #include "mlx5_flow.h" #include "mlx5_flow_os.h" -#include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_common_os.h" #include "rte_pmd_mlx5.h" diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c index cac05fb..cb5b3c9 100644 --- a/drivers/net/mlx5/mlx5_flow_dv.c +++ b/drivers/net/mlx5/mlx5_flow_dv.c @@ -32,8 +32,8 @@ #include "mlx5_common_os.h" #include "mlx5_flow.h" #include "mlx5_flow_os.h" -#include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "rte_pmd_mlx5.h" #if defined(HAVE_IBV_FLOW_DV_SUPPORT) || !defined(HAVE_INFINIBAND_VERBS_H) diff --git a/drivers/net/mlx5/mlx5_flow_verbs.c b/drivers/net/mlx5/mlx5_flow_verbs.c index c331350..0fdafbb 100644 --- a/drivers/net/mlx5/mlx5_flow_verbs.c +++ b/drivers/net/mlx5/mlx5_flow_verbs.c @@ -23,7 +23,6 @@ #include "mlx5_defs.h" #include "mlx5.h" #include "mlx5_flow.h" -#include "mlx5_rxtx.h" #include "mlx5_rx.h" #define VERBS_SPEC_INNER(item_flags) \ diff --git a/drivers/net/mlx5/mlx5_mr.c b/drivers/net/mlx5/mlx5_mr.c index 2014936..e791b63 100644 --- a/drivers/net/mlx5/mlx5_mr.c +++ b/drivers/net/mlx5/mlx5_mr.c @@ -16,6 +16,7 @@ #include "mlx5_mr.h" #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" struct mr_find_contig_memsegs_data { uintptr_t addr; diff --git a/drivers/net/mlx5/mlx5_rxmode.c b/drivers/net/mlx5/mlx5_rxmode.c index cf93cca..25fb47c 100644 --- a/drivers/net/mlx5/mlx5_rxmode.c +++ b/drivers/net/mlx5/mlx5_rxmode.c @@ -11,7 +11,6 @@ #include #include "mlx5.h" -#include "mlx5_rxtx.h" #include "mlx5_utils.h" /** diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c index 19df0fa..bb9a908 100644 --- a/drivers/net/mlx5/mlx5_rxq.c +++ b/drivers/net/mlx5/mlx5_rxq.c @@ -24,7 +24,7 @@ #include "mlx5_defs.h" #include "mlx5.h" -#include "mlx5_rxtx.h" +#include "mlx5_tx.h" #include "mlx5_rx.h" #include "mlx5_utils.h" #include "mlx5_autoconf.h" diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c index c7f2605..57ff407 100644 --- a/drivers/net/mlx5/mlx5_rxtx.c +++ b/drivers/net/mlx5/mlx5_rxtx.c @@ -26,6 +26,7 @@ #include "mlx5_utils.h" #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" /* TX burst subroutines return codes. */ enum mlx5_txcmp_code { diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h index f1ebc99..e168dd4 100644 --- a/drivers/net/mlx5/mlx5_rxtx.h +++ b/drivers/net/mlx5/mlx5_rxtx.h @@ -17,169 +17,18 @@ #include #include #include -#include #include -#include -#include #include #include -#include "mlx5_defs.h" #include "mlx5_utils.h" #include "mlx5.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" - -/* Mbuf dynamic flag offset for inline. */ -extern uint64_t rte_net_mlx5_dynf_inline_mask; - -struct mlx5_txq_stats { -#ifdef MLX5_PMD_SOFT_COUNTERS - uint64_t opackets; /**< Total of successfully sent packets. */ - uint64_t obytes; /**< Total of successfully sent bytes. */ -#endif - uint64_t oerrors; /**< Total number of failed transmitted packets. */ -}; - struct mlx5_priv; -/* TX queue send local data. */ -__extension__ -struct mlx5_txq_local { - struct mlx5_wqe *wqe_last; /* last sent WQE pointer. */ - struct rte_mbuf *mbuf; /* first mbuf to process. */ - uint16_t pkts_copy; /* packets copied to elts. */ - uint16_t pkts_sent; /* packets sent. */ - uint16_t pkts_loop; /* packets sent on loop entry. */ - uint16_t elts_free; /* available elts remain. */ - uint16_t wqe_free; /* available wqe remain. */ - uint16_t mbuf_off; /* data offset in current mbuf. */ - uint16_t mbuf_nseg; /* number of remaining mbuf. */ - uint16_t mbuf_free; /* number of inline mbufs to free. */ -}; - -/* TX queue descriptor. */ -__extension__ -struct mlx5_txq_data { - uint16_t elts_head; /* Current counter in (*elts)[]. */ - uint16_t elts_tail; /* Counter of first element awaiting completion. */ - uint16_t elts_comp; /* elts index since last completion request. */ - uint16_t elts_s; /* Number of mbuf elements. */ - uint16_t elts_m; /* Mask for mbuf elements indices. */ - /* Fields related to elts mbuf storage. */ - uint16_t wqe_ci; /* Consumer index for work queue. */ - uint16_t wqe_pi; /* Producer index for work queue. */ - uint16_t wqe_s; /* Number of WQ elements. */ - uint16_t wqe_m; /* Mask Number for WQ elements. */ - uint16_t wqe_comp; /* WQE index since last completion request. */ - uint16_t wqe_thres; /* WQE threshold to request completion in CQ. */ - /* WQ related fields. */ - uint16_t cq_ci; /* Consumer index for completion queue. */ - uint16_t cq_pi; /* Production index for completion queue. */ - uint16_t cqe_s; /* Number of CQ elements. */ - uint16_t cqe_m; /* Mask for CQ indices. */ - /* CQ related fields. */ - uint16_t elts_n:4; /* elts[] length (in log2). */ - uint16_t cqe_n:4; /* Number of CQ elements (in log2). */ - uint16_t wqe_n:4; /* Number of WQ elements (in log2). */ - uint16_t tso_en:1; /* When set hardware TSO is enabled. */ - uint16_t tunnel_en:1; - /* When set TX offload for tunneled packets are supported. */ - uint16_t swp_en:1; /* Whether SW parser is enabled. */ - uint16_t vlan_en:1; /* VLAN insertion in WQE is supported. */ - uint16_t db_nc:1; /* Doorbell mapped to non-cached region. */ - uint16_t db_heu:1; /* Doorbell heuristic write barrier. */ - uint16_t fast_free:1; /* mbuf fast free on Tx is enabled. */ - uint16_t inlen_send; /* Ordinary send data inline size. */ - uint16_t inlen_empw; /* eMPW max packet size to inline. */ - uint16_t inlen_mode; /* Minimal data length to inline. */ - uint32_t qp_num_8s; /* QP number shifted by 8. */ - uint64_t offloads; /* Offloads for Tx Queue. */ - struct mlx5_mr_ctrl mr_ctrl; /* MR control descriptor. */ - struct mlx5_wqe *wqes; /* Work queue. */ - struct mlx5_wqe *wqes_end; /* Work queue array limit. */ -#ifdef RTE_LIBRTE_MLX5_DEBUG - uint32_t *fcqs; /* Free completion queue (debug extended). */ -#else - uint16_t *fcqs; /* Free completion queue. */ -#endif - volatile struct mlx5_cqe *cqes; /* Completion queue. */ - volatile uint32_t *qp_db; /* Work queue doorbell. */ - volatile uint32_t *cq_db; /* Completion queue doorbell. */ - uint16_t port_id; /* Port ID of device. */ - uint16_t idx; /* Queue index. */ - uint64_t ts_mask; /* Timestamp flag dynamic mask. */ - int32_t ts_offset; /* Timestamp field dynamic offset. */ - struct mlx5_dev_ctx_shared *sh; /* Shared context. */ - struct mlx5_txq_stats stats; /* TX queue counters. */ -#ifndef RTE_ARCH_64 - rte_spinlock_t *uar_lock; - /* UAR access lock required for 32bit implementations */ -#endif - struct rte_mbuf *elts[0]; - /* Storage for queued packets, must be the last field. */ -} __rte_cache_aligned; - -enum mlx5_txq_type { - MLX5_TXQ_TYPE_STANDARD, /* Standard Tx queue. */ - MLX5_TXQ_TYPE_HAIRPIN, /* Hairpin Rx queue. */ -}; - -/* TX queue control descriptor. */ -struct mlx5_txq_ctrl { - LIST_ENTRY(mlx5_txq_ctrl) next; /* Pointer to the next element. */ - uint32_t refcnt; /* Reference counter. */ - unsigned int socket; /* CPU socket ID for allocations. */ - enum mlx5_txq_type type; /* The txq ctrl type. */ - unsigned int max_inline_data; /* Max inline data. */ - unsigned int max_tso_header; /* Max TSO header size. */ - struct mlx5_txq_obj *obj; /* Verbs/DevX queue object. */ - struct mlx5_priv *priv; /* Back pointer to private data. */ - off_t uar_mmap_offset; /* UAR mmap offset for non-primary process. */ - void *bf_reg; /* BlueFlame register from Verbs. */ - uint16_t dump_file_n; /* Number of dump files. */ - struct rte_eth_hairpin_conf hairpin_conf; /* Hairpin configuration. */ - uint32_t hairpin_status; /* Hairpin binding status. */ - struct mlx5_txq_data txq; /* Data path structure. */ - /* Must be the last field in the structure, contains elts[]. */ -}; - -#define MLX5_TX_BFREG(txq) \ - (MLX5_PROC_PRIV((txq)->port_id)->uar_table[(txq)->idx]) - -/* mlx5_txq.c */ - -int mlx5_tx_queue_start(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_tx_queue_stop(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_tx_queue_start_primary(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_tx_queue_stop_primary(struct rte_eth_dev *dev, uint16_t queue_id); -int mlx5_tx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, - unsigned int socket, const struct rte_eth_txconf *conf); -int mlx5_tx_hairpin_queue_setup - (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, - const struct rte_eth_hairpin_conf *hairpin_conf); -void mlx5_tx_queue_release(void *dpdk_txq); -void txq_uar_init(struct mlx5_txq_ctrl *txq_ctrl); -int mlx5_tx_uar_init_secondary(struct rte_eth_dev *dev, int fd); -void mlx5_tx_uar_uninit_secondary(struct rte_eth_dev *dev); -int mlx5_txq_obj_verify(struct rte_eth_dev *dev); -struct mlx5_txq_ctrl *mlx5_txq_new(struct rte_eth_dev *dev, uint16_t idx, - uint16_t desc, unsigned int socket, - const struct rte_eth_txconf *conf); -struct mlx5_txq_ctrl *mlx5_txq_hairpin_new - (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, - const struct rte_eth_hairpin_conf *hairpin_conf); -struct mlx5_txq_ctrl *mlx5_txq_get(struct rte_eth_dev *dev, uint16_t idx); -int mlx5_txq_release(struct rte_eth_dev *dev, uint16_t idx); -int mlx5_txq_releasable(struct rte_eth_dev *dev, uint16_t idx); -int mlx5_txq_verify(struct rte_eth_dev *dev); -void txq_alloc_elts(struct mlx5_txq_ctrl *txq_ctrl); -void txq_free_elts(struct mlx5_txq_ctrl *txq_ctrl); -uint64_t mlx5_get_tx_port_offloads(struct rte_eth_dev *dev); -void mlx5_txq_dynf_timestamp_set(struct rte_eth_dev *dev); - /* mlx5_rxtx.c */ extern uint32_t mlx5_ptype_table[]; @@ -189,88 +38,22 @@ struct mlx5_txq_ctrl *mlx5_txq_hairpin_new void mlx5_set_ptype_table(void); void mlx5_set_cksum_table(void); void mlx5_set_swp_types_table(void); -uint16_t removed_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, - uint16_t pkts_n); -int mlx5_tx_descriptor_status(void *tx_queue, uint16_t offset); void mlx5_dump_debug_information(const char *path, const char *title, const void *buf, unsigned int len); int mlx5_queue_state_modify_primary(struct rte_eth_dev *dev, const struct mlx5_mp_arg_queue_state_modify *sm); int mlx5_queue_state_modify(struct rte_eth_dev *dev, struct mlx5_mp_arg_queue_state_modify *sm); -void mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, - struct rte_eth_txq_info *qinfo); -int mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, - struct rte_eth_burst_mode *mode); /* mlx5_mr.c */ void mlx5_mr_flush_local_cache(struct mlx5_mr_ctrl *mr_ctrl); -uint32_t mlx5_tx_mb2mr_bh(struct mlx5_txq_data *txq, struct rte_mbuf *mb); -uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, - struct rte_mempool *mp); int mlx5_dma_map(struct rte_pci_device *pdev, void *addr, uint64_t iova, size_t len); int mlx5_dma_unmap(struct rte_pci_device *pdev, void *addr, uint64_t iova, size_t len); /** - * Provide safe 64bit store operation to mlx5 UAR region for both 32bit and - * 64bit architectures. - * - * @param val - * value to write in CPU endian format. - * @param addr - * Address to write to. - * @param lock - * Address of the lock to use for that UAR access. - */ -static __rte_always_inline void -__mlx5_uar_write64_relaxed(uint64_t val, void *addr, - rte_spinlock_t *lock __rte_unused) -{ -#ifdef RTE_ARCH_64 - *(uint64_t *)addr = val; -#else /* !RTE_ARCH_64 */ - rte_spinlock_lock(lock); - *(uint32_t *)addr = val; - rte_io_wmb(); - *((uint32_t *)addr + 1) = val >> 32; - rte_spinlock_unlock(lock); -#endif -} - -/** - * Provide safe 64bit store operation to mlx5 UAR region for both 32bit and - * 64bit architectures while guaranteeing the order of execution with the - * code being executed. - * - * @param val - * value to write in CPU endian format. - * @param addr - * Address to write to. - * @param lock - * Address of the lock to use for that UAR access. - */ -static __rte_always_inline void -__mlx5_uar_write64(uint64_t val, void *addr, rte_spinlock_t *lock) -{ - rte_io_wmb(); - __mlx5_uar_write64_relaxed(val, addr, lock); -} - -/* Assist macros, used instead of directly calling the functions they wrap. */ -#ifdef RTE_ARCH_64 -#define mlx5_uar_write64_relaxed(val, dst, lock) \ - __mlx5_uar_write64_relaxed(val, dst, NULL) -#define mlx5_uar_write64(val, dst, lock) __mlx5_uar_write64(val, dst, NULL) -#else -#define mlx5_uar_write64_relaxed(val, dst, lock) \ - __mlx5_uar_write64_relaxed(val, dst, lock) -#define mlx5_uar_write64(val, dst, lock) __mlx5_uar_write64(val, dst, lock) -#endif - -/** * Get Memory Pool (MP) from mbuf. If mbuf is indirect, the pool from which the * cloned mbuf is allocated is returned instead. * @@ -288,131 +71,4 @@ int mlx5_dma_unmap(struct rte_pci_device *pdev, void *addr, uint64_t iova, return buf->pool; } -/** - * Query LKey from a packet buffer for Tx. If not found, add the mempool. - * - * @param txq - * Pointer to Tx queue structure. - * @param addr - * Address to search. - * - * @return - * Searched LKey on success, UINT32_MAX on no match. - */ -static __rte_always_inline uint32_t -mlx5_tx_mb2mr(struct mlx5_txq_data *txq, struct rte_mbuf *mb) -{ - struct mlx5_mr_ctrl *mr_ctrl = &txq->mr_ctrl; - uintptr_t addr = (uintptr_t)mb->buf_addr; - uint32_t lkey; - - /* Check generation bit to see if there's any change on existing MRs. */ - if (unlikely(*mr_ctrl->dev_gen_ptr != mr_ctrl->cur_gen)) - mlx5_mr_flush_local_cache(mr_ctrl); - /* Linear search on MR cache array. */ - lkey = mlx5_mr_lookup_lkey(mr_ctrl->cache, &mr_ctrl->mru, - MLX5_MR_CACHE_N, addr); - if (likely(lkey != UINT32_MAX)) - return lkey; - /* Take slower bottom-half on miss. */ - return mlx5_tx_mb2mr_bh(txq, mb); -} - -/** - * Ring TX queue doorbell and flush the update if requested. - * - * @param txq - * Pointer to TX queue structure. - * @param wqe - * Pointer to the last WQE posted in the NIC. - * @param cond - * Request for write memory barrier after BlueFlame update. - */ -static __rte_always_inline void -mlx5_tx_dbrec_cond_wmb(struct mlx5_txq_data *txq, volatile struct mlx5_wqe *wqe, - int cond) -{ - uint64_t *dst = MLX5_TX_BFREG(txq); - volatile uint64_t *src = ((volatile uint64_t *)wqe); - - rte_io_wmb(); - *txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci); - /* Ensure ordering between DB record and BF copy. */ - rte_wmb(); - mlx5_uar_write64_relaxed(*src, dst, txq->uar_lock); - if (cond) - rte_wmb(); -} - -/** - * Ring TX queue doorbell and flush the update by write memory barrier. - * - * @param txq - * Pointer to TX queue structure. - * @param wqe - * Pointer to the last WQE posted in the NIC. - */ -static __rte_always_inline void -mlx5_tx_dbrec(struct mlx5_txq_data *txq, volatile struct mlx5_wqe *wqe) -{ - mlx5_tx_dbrec_cond_wmb(txq, wqe, 1); -} - -/** - * Convert timestamp from mbuf format to linear counter - * of Clock Queue completions (24 bits) - * - * @param sh - * Pointer to the device shared context to fetch Tx - * packet pacing timestamp and parameters. - * @param ts - * Timestamp from mbuf to convert. - * @return - * positive or zero value - completion ID to wait - * negative value - conversion error - */ -static __rte_always_inline int32_t -mlx5_txpp_convert_tx_ts(struct mlx5_dev_ctx_shared *sh, uint64_t mts) -{ - uint64_t ts, ci; - uint32_t tick; - - do { - /* - * Read atomically two uint64_t fields and compare lsb bits. - * It there is no match - the timestamp was updated in - * the service thread, data should be re-read. - */ - rte_compiler_barrier(); - ci = __atomic_load_n(&sh->txpp.ts.ci_ts, __ATOMIC_RELAXED); - ts = __atomic_load_n(&sh->txpp.ts.ts, __ATOMIC_RELAXED); - rte_compiler_barrier(); - if (!((ts ^ ci) << (64 - MLX5_CQ_INDEX_WIDTH))) - break; - } while (true); - /* Perform the skew correction, positive value to send earlier. */ - mts -= sh->txpp.skew; - mts -= ts; - if (unlikely(mts >= UINT64_MAX / 2)) { - /* We have negative integer, mts is in the past. */ - __atomic_fetch_add(&sh->txpp.err_ts_past, - 1, __ATOMIC_RELAXED); - return -1; - } - tick = sh->txpp.tick; - MLX5_ASSERT(tick); - /* Convert delta to completions, round up. */ - mts = (mts + tick - 1) / tick; - if (unlikely(mts >= (1 << MLX5_CQ_INDEX_WIDTH) / 2 - 1)) { - /* We have mts is too distant future. */ - __atomic_fetch_add(&sh->txpp.err_ts_future, - 1, __ATOMIC_RELAXED); - return -1; - } - mts <<= 64 - MLX5_CQ_INDEX_WIDTH; - ci += mts; - ci >>= 64 - MLX5_CQ_INDEX_WIDTH; - return ci; -} - #endif /* RTE_PMD_MLX5_RXTX_H_ */ diff --git a/drivers/net/mlx5/mlx5_stats.c b/drivers/net/mlx5/mlx5_stats.c index 4dbd831..ae2f566 100644 --- a/drivers/net/mlx5/mlx5_stats.c +++ b/drivers/net/mlx5/mlx5_stats.c @@ -16,8 +16,8 @@ #include "mlx5_defs.h" #include "mlx5.h" -#include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_malloc.h" /** diff --git a/drivers/net/mlx5/mlx5_trigger.c b/drivers/net/mlx5/mlx5_trigger.c index c88cb22..001c0b5 100644 --- a/drivers/net/mlx5/mlx5_trigger.c +++ b/drivers/net/mlx5/mlx5_trigger.c @@ -15,8 +15,8 @@ #include "mlx5.h" #include "mlx5_mr.h" -#include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_utils.h" #include "rte_pmd_mlx5.h" diff --git a/drivers/net/mlx5/mlx5_tx.h b/drivers/net/mlx5/mlx5_tx.h new file mode 100644 index 0000000..7f91d04 --- /dev/null +++ b/drivers/net/mlx5/mlx5_tx.h @@ -0,0 +1,371 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#ifndef RTE_PMD_MLX5_TX_H_ +#define RTE_PMD_MLX5_TX_H_ + +#include +#include + +#include +#include +#include +#include + +#include + +#include "mlx5.h" +#include "mlx5_autoconf.h" +#include "mlx5_mr.h" + +/* Mbuf dynamic flag offset for inline. */ +extern uint64_t rte_net_mlx5_dynf_inline_mask; + +struct mlx5_txq_stats { +#ifdef MLX5_PMD_SOFT_COUNTERS + uint64_t opackets; /**< Total of successfully sent packets. */ + uint64_t obytes; /**< Total of successfully sent bytes. */ +#endif + uint64_t oerrors; /**< Total number of failed transmitted packets. */ +}; + +/* TX queue send local data. */ +__extension__ +struct mlx5_txq_local { + struct mlx5_wqe *wqe_last; /* last sent WQE pointer. */ + struct rte_mbuf *mbuf; /* first mbuf to process. */ + uint16_t pkts_copy; /* packets copied to elts. */ + uint16_t pkts_sent; /* packets sent. */ + uint16_t pkts_loop; /* packets sent on loop entry. */ + uint16_t elts_free; /* available elts remain. */ + uint16_t wqe_free; /* available wqe remain. */ + uint16_t mbuf_off; /* data offset in current mbuf. */ + uint16_t mbuf_nseg; /* number of remaining mbuf. */ + uint16_t mbuf_free; /* number of inline mbufs to free. */ +}; + +/* TX queue descriptor. */ +__extension__ +struct mlx5_txq_data { + uint16_t elts_head; /* Current counter in (*elts)[]. */ + uint16_t elts_tail; /* Counter of first element awaiting completion. */ + uint16_t elts_comp; /* elts index since last completion request. */ + uint16_t elts_s; /* Number of mbuf elements. */ + uint16_t elts_m; /* Mask for mbuf elements indices. */ + /* Fields related to elts mbuf storage. */ + uint16_t wqe_ci; /* Consumer index for work queue. */ + uint16_t wqe_pi; /* Producer index for work queue. */ + uint16_t wqe_s; /* Number of WQ elements. */ + uint16_t wqe_m; /* Mask Number for WQ elements. */ + uint16_t wqe_comp; /* WQE index since last completion request. */ + uint16_t wqe_thres; /* WQE threshold to request completion in CQ. */ + /* WQ related fields. */ + uint16_t cq_ci; /* Consumer index for completion queue. */ + uint16_t cq_pi; /* Production index for completion queue. */ + uint16_t cqe_s; /* Number of CQ elements. */ + uint16_t cqe_m; /* Mask for CQ indices. */ + /* CQ related fields. */ + uint16_t elts_n:4; /* elts[] length (in log2). */ + uint16_t cqe_n:4; /* Number of CQ elements (in log2). */ + uint16_t wqe_n:4; /* Number of WQ elements (in log2). */ + uint16_t tso_en:1; /* When set hardware TSO is enabled. */ + uint16_t tunnel_en:1; + /* When set TX offload for tunneled packets are supported. */ + uint16_t swp_en:1; /* Whether SW parser is enabled. */ + uint16_t vlan_en:1; /* VLAN insertion in WQE is supported. */ + uint16_t db_nc:1; /* Doorbell mapped to non-cached region. */ + uint16_t db_heu:1; /* Doorbell heuristic write barrier. */ + uint16_t fast_free:1; /* mbuf fast free on Tx is enabled. */ + uint16_t inlen_send; /* Ordinary send data inline size. */ + uint16_t inlen_empw; /* eMPW max packet size to inline. */ + uint16_t inlen_mode; /* Minimal data length to inline. */ + uint32_t qp_num_8s; /* QP number shifted by 8. */ + uint64_t offloads; /* Offloads for Tx Queue. */ + struct mlx5_mr_ctrl mr_ctrl; /* MR control descriptor. */ + struct mlx5_wqe *wqes; /* Work queue. */ + struct mlx5_wqe *wqes_end; /* Work queue array limit. */ +#ifdef RTE_LIBRTE_MLX5_DEBUG + uint32_t *fcqs; /* Free completion queue (debug extended). */ +#else + uint16_t *fcqs; /* Free completion queue. */ +#endif + volatile struct mlx5_cqe *cqes; /* Completion queue. */ + volatile uint32_t *qp_db; /* Work queue doorbell. */ + volatile uint32_t *cq_db; /* Completion queue doorbell. */ + uint16_t port_id; /* Port ID of device. */ + uint16_t idx; /* Queue index. */ + uint64_t ts_mask; /* Timestamp flag dynamic mask. */ + int32_t ts_offset; /* Timestamp field dynamic offset. */ + struct mlx5_dev_ctx_shared *sh; /* Shared context. */ + struct mlx5_txq_stats stats; /* TX queue counters. */ +#ifndef RTE_ARCH_64 + rte_spinlock_t *uar_lock; + /* UAR access lock required for 32bit implementations */ +#endif + struct rte_mbuf *elts[0]; + /* Storage for queued packets, must be the last field. */ +} __rte_cache_aligned; + +enum mlx5_txq_type { + MLX5_TXQ_TYPE_STANDARD, /* Standard Tx queue. */ + MLX5_TXQ_TYPE_HAIRPIN, /* Hairpin Tx queue. */ +}; + +/* TX queue control descriptor. */ +struct mlx5_txq_ctrl { + LIST_ENTRY(mlx5_txq_ctrl) next; /* Pointer to the next element. */ + uint32_t refcnt; /* Reference counter. */ + unsigned int socket; /* CPU socket ID for allocations. */ + enum mlx5_txq_type type; /* The txq ctrl type. */ + unsigned int max_inline_data; /* Max inline data. */ + unsigned int max_tso_header; /* Max TSO header size. */ + struct mlx5_txq_obj *obj; /* Verbs/DevX queue object. */ + struct mlx5_priv *priv; /* Back pointer to private data. */ + off_t uar_mmap_offset; /* UAR mmap offset for non-primary process. */ + void *bf_reg; /* BlueFlame register from Verbs. */ + uint16_t dump_file_n; /* Number of dump files. */ + struct rte_eth_hairpin_conf hairpin_conf; /* Hairpin configuration. */ + uint32_t hairpin_status; /* Hairpin binding status. */ + struct mlx5_txq_data txq; /* Data path structure. */ + /* Must be the last field in the structure, contains elts[]. */ +}; + +/* mlx5_txq.c */ + +int mlx5_tx_queue_start(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_tx_queue_stop(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_tx_queue_start_primary(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_tx_queue_stop_primary(struct rte_eth_dev *dev, uint16_t queue_id); +int mlx5_tx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, + unsigned int socket, const struct rte_eth_txconf *conf); +int mlx5_tx_hairpin_queue_setup + (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, + const struct rte_eth_hairpin_conf *hairpin_conf); +void mlx5_tx_queue_release(void *dpdk_txq); +void txq_uar_init(struct mlx5_txq_ctrl *txq_ctrl); +int mlx5_tx_uar_init_secondary(struct rte_eth_dev *dev, int fd); +void mlx5_tx_uar_uninit_secondary(struct rte_eth_dev *dev); +int mlx5_txq_obj_verify(struct rte_eth_dev *dev); +struct mlx5_txq_ctrl *mlx5_txq_new(struct rte_eth_dev *dev, uint16_t idx, + uint16_t desc, unsigned int socket, + const struct rte_eth_txconf *conf); +struct mlx5_txq_ctrl *mlx5_txq_hairpin_new + (struct rte_eth_dev *dev, uint16_t idx, uint16_t desc, + const struct rte_eth_hairpin_conf *hairpin_conf); +struct mlx5_txq_ctrl *mlx5_txq_get(struct rte_eth_dev *dev, uint16_t idx); +int mlx5_txq_release(struct rte_eth_dev *dev, uint16_t idx); +int mlx5_txq_releasable(struct rte_eth_dev *dev, uint16_t idx); +int mlx5_txq_verify(struct rte_eth_dev *dev); +void txq_alloc_elts(struct mlx5_txq_ctrl *txq_ctrl); +void txq_free_elts(struct mlx5_txq_ctrl *txq_ctrl); +uint64_t mlx5_get_tx_port_offloads(struct rte_eth_dev *dev); +void mlx5_txq_dynf_timestamp_set(struct rte_eth_dev *dev); + +/* mlx5_tx.c */ + +uint16_t removed_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, + uint16_t pkts_n); +int mlx5_tx_descriptor_status(void *tx_queue, uint16_t offset); +void mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, + struct rte_eth_txq_info *qinfo); +int mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, + struct rte_eth_burst_mode *mode); + +/* mlx5_mr.c */ + +uint32_t mlx5_tx_mb2mr_bh(struct mlx5_txq_data *txq, struct rte_mbuf *mb); +uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, + struct rte_mempool *mp); + +static __rte_always_inline uint64_t * +mlx5_tx_bfreg(struct mlx5_txq_data *txq) +{ + return MLX5_PROC_PRIV(txq->port_id)->uar_table[txq->idx]; +} + +/** + * Provide safe 64bit store operation to mlx5 UAR region for both 32bit and + * 64bit architectures. + * + * @param val + * value to write in CPU endian format. + * @param addr + * Address to write to. + * @param lock + * Address of the lock to use for that UAR access. + */ +static __rte_always_inline void +__mlx5_uar_write64_relaxed(uint64_t val, void *addr, + rte_spinlock_t *lock __rte_unused) +{ +#ifdef RTE_ARCH_64 + *(uint64_t *)addr = val; +#else /* !RTE_ARCH_64 */ + rte_spinlock_lock(lock); + *(uint32_t *)addr = val; + rte_io_wmb(); + *((uint32_t *)addr + 1) = val >> 32; + rte_spinlock_unlock(lock); +#endif +} + +/** + * Provide safe 64bit store operation to mlx5 UAR region for both 32bit and + * 64bit architectures while guaranteeing the order of execution with the + * code being executed. + * + * @param val + * value to write in CPU endian format. + * @param addr + * Address to write to. + * @param lock + * Address of the lock to use for that UAR access. + */ +static __rte_always_inline void +__mlx5_uar_write64(uint64_t val, void *addr, rte_spinlock_t *lock) +{ + rte_io_wmb(); + __mlx5_uar_write64_relaxed(val, addr, lock); +} + +/* Assist macros, used instead of directly calling the functions they wrap. */ +#ifdef RTE_ARCH_64 +#define mlx5_uar_write64_relaxed(val, dst, lock) \ + __mlx5_uar_write64_relaxed(val, dst, NULL) +#define mlx5_uar_write64(val, dst, lock) __mlx5_uar_write64(val, dst, NULL) +#else +#define mlx5_uar_write64_relaxed(val, dst, lock) \ + __mlx5_uar_write64_relaxed(val, dst, lock) +#define mlx5_uar_write64(val, dst, lock) __mlx5_uar_write64(val, dst, lock) +#endif + +/** + * Query LKey from a packet buffer for Tx. If not found, add the mempool. + * + * @param txq + * Pointer to Tx queue structure. + * @param addr + * Address to search. + * + * @return + * Searched LKey on success, UINT32_MAX on no match. + */ +static __rte_always_inline uint32_t +mlx5_tx_mb2mr(struct mlx5_txq_data *txq, struct rte_mbuf *mb) +{ + struct mlx5_mr_ctrl *mr_ctrl = &txq->mr_ctrl; + uintptr_t addr = (uintptr_t)mb->buf_addr; + uint32_t lkey; + + /* Check generation bit to see if there's any change on existing MRs. */ + if (unlikely(*mr_ctrl->dev_gen_ptr != mr_ctrl->cur_gen)) + mlx5_mr_flush_local_cache(mr_ctrl); + /* Linear search on MR cache array. */ + lkey = mlx5_mr_lookup_lkey(mr_ctrl->cache, &mr_ctrl->mru, + MLX5_MR_CACHE_N, addr); + if (likely(lkey != UINT32_MAX)) + return lkey; + /* Take slower bottom-half on miss. */ + return mlx5_tx_mb2mr_bh(txq, mb); +} + +/** + * Ring TX queue doorbell and flush the update if requested. + * + * @param txq + * Pointer to TX queue structure. + * @param wqe + * Pointer to the last WQE posted in the NIC. + * @param cond + * Request for write memory barrier after BlueFlame update. + */ +static __rte_always_inline void +mlx5_tx_dbrec_cond_wmb(struct mlx5_txq_data *txq, volatile struct mlx5_wqe *wqe, + int cond) +{ + uint64_t *dst = mlx5_tx_bfreg(txq); + volatile uint64_t *src = ((volatile uint64_t *)wqe); + + rte_io_wmb(); + *txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci); + /* Ensure ordering between DB record and BF copy. */ + rte_wmb(); + mlx5_uar_write64_relaxed(*src, dst, txq->uar_lock); + if (cond) + rte_wmb(); +} + +/** + * Ring TX queue doorbell and flush the update by write memory barrier. + * + * @param txq + * Pointer to TX queue structure. + * @param wqe + * Pointer to the last WQE posted in the NIC. + */ +static __rte_always_inline void +mlx5_tx_dbrec(struct mlx5_txq_data *txq, volatile struct mlx5_wqe *wqe) +{ + mlx5_tx_dbrec_cond_wmb(txq, wqe, 1); +} + +/** + * Convert timestamp from mbuf format to linear counter + * of Clock Queue completions (24 bits). + * + * @param sh + * Pointer to the device shared context to fetch Tx + * packet pacing timestamp and parameters. + * @param ts + * Timestamp from mbuf to convert. + * @return + * positive or zero value - completion ID to wait. + * negative value - conversion error. + */ +static __rte_always_inline int32_t +mlx5_txpp_convert_tx_ts(struct mlx5_dev_ctx_shared *sh, uint64_t mts) +{ + uint64_t ts, ci; + uint32_t tick; + + do { + /* + * Read atomically two uint64_t fields and compare lsb bits. + * It there is no match - the timestamp was updated in + * the service thread, data should be re-read. + */ + rte_compiler_barrier(); + ci = __atomic_load_n(&sh->txpp.ts.ci_ts, __ATOMIC_RELAXED); + ts = __atomic_load_n(&sh->txpp.ts.ts, __ATOMIC_RELAXED); + rte_compiler_barrier(); + if (!((ts ^ ci) << (64 - MLX5_CQ_INDEX_WIDTH))) + break; + } while (true); + /* Perform the skew correction, positive value to send earlier. */ + mts -= sh->txpp.skew; + mts -= ts; + if (unlikely(mts >= UINT64_MAX / 2)) { + /* We have negative integer, mts is in the past. */ + __atomic_fetch_add(&sh->txpp.err_ts_past, + 1, __ATOMIC_RELAXED); + return -1; + } + tick = sh->txpp.tick; + MLX5_ASSERT(tick); + /* Convert delta to completions, round up. */ + mts = (mts + tick - 1) / tick; + if (unlikely(mts >= (1 << MLX5_CQ_INDEX_WIDTH) / 2 - 1)) { + /* We have mts is too distant future. */ + __atomic_fetch_add(&sh->txpp.err_ts_future, + 1, __ATOMIC_RELAXED); + return -1; + } + mts <<= 64 - MLX5_CQ_INDEX_WIDTH; + ci += mts; + ci >>= 64 - MLX5_CQ_INDEX_WIDTH; + return ci; +} + +#endif /* RTE_PMD_MLX5_TX_H_ */ diff --git a/drivers/net/mlx5/mlx5_txpp.c b/drivers/net/mlx5/mlx5_txpp.c index 89e1c5d..d90399a 100644 --- a/drivers/net/mlx5/mlx5_txpp.c +++ b/drivers/net/mlx5/mlx5_txpp.c @@ -16,8 +16,8 @@ #include #include "mlx5.h" -#include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_common_os.h" static_assert(sizeof(struct mlx5_cqe_ts) == sizeof(rte_int128_t), diff --git a/drivers/net/mlx5/mlx5_txq.c b/drivers/net/mlx5/mlx5_txq.c index cd13eb9..b8a1657 100644 --- a/drivers/net/mlx5/mlx5_txq.c +++ b/drivers/net/mlx5/mlx5_txq.c @@ -23,7 +23,7 @@ #include "mlx5_defs.h" #include "mlx5_utils.h" #include "mlx5.h" -#include "mlx5_rxtx.h" +#include "mlx5_tx.h" #include "mlx5_autoconf.h" /** diff --git a/drivers/net/mlx5/windows/mlx5_os.c b/drivers/net/mlx5/windows/mlx5_os.c index 79eac80..814063b 100644 --- a/drivers/net/mlx5/windows/mlx5_os.c +++ b/drivers/net/mlx5/windows/mlx5_os.c @@ -24,6 +24,7 @@ #include "mlx5_utils.h" #include "mlx5_rxtx.h" #include "mlx5_rx.h" +#include "mlx5_tx.h" #include "mlx5_autoconf.h" #include "mlx5_mr.h" #include "mlx5_flow.h" From patchwork Mon Apr 5 14:00:54 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Baum X-Patchwork-Id: 90564 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id 80DD2A0A02; Mon, 5 Apr 2021 16:02:36 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id EF6B7140F38; Mon, 5 Apr 2021 16:01:43 +0200 (CEST) Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by mails.dpdk.org (Postfix) with ESMTP id A387E140EAA for ; Mon, 5 Apr 2021 16:01:35 +0200 (CEST) Received: from Internal Mail-Server by MTLPINE1 (envelope-from michaelba@nvidia.com) with SMTP; 5 Apr 2021 17:01:32 +0300 Received: from nvidia.com (pegasus07.mtr.labs.mlnx [10.210.16.112]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 135E1VNa031476; Mon, 5 Apr 2021 17:01:32 +0300 From: Michael Baum To: dev@dpdk.org Cc: Matan Azrad , Raslan Darawsheh , Viacheslav Ovsiienko Date: Mon, 5 Apr 2021 14:00:54 +0000 Message-Id: <1617631256-3018-5-git-send-email-michaelba@nvidia.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> References: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> Subject: [dpdk-dev] [PATCH 4/6] net/mlx5: separate Tx burst template to header file X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" This patch moves Tx burst and its inline functions declarations to header file to allow its use from several separate source files and as a possible preparation for Tx cleanup. Signed-off-by: Michael Baum Acked-by: Viacheslav Ovsiienko --- drivers/net/mlx5/mlx5_rxtx.c | 3331 +----------------------------------------- drivers/net/mlx5/mlx5_tx.h | 3304 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 3305 insertions(+), 3330 deletions(-) diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c index 57ff407..2f36754 100644 --- a/drivers/net/mlx5/mlx5_rxtx.c +++ b/drivers/net/mlx5/mlx5_rxtx.c @@ -28,57 +28,6 @@ #include "mlx5_rx.h" #include "mlx5_tx.h" -/* TX burst subroutines return codes. */ -enum mlx5_txcmp_code { - MLX5_TXCMP_CODE_EXIT = 0, - MLX5_TXCMP_CODE_ERROR, - MLX5_TXCMP_CODE_SINGLE, - MLX5_TXCMP_CODE_MULTI, - MLX5_TXCMP_CODE_TSO, - MLX5_TXCMP_CODE_EMPW, -}; - -/* - * These defines are used to configure Tx burst routine option set - * supported at compile time. The not specified options are optimized out - * out due to if conditions can be explicitly calculated at compile time. - * The offloads with bigger runtime check (require more CPU cycles to - * skip) overhead should have the bigger index - this is needed to - * select the better matching routine function if no exact match and - * some offloads are not actually requested. - */ -#define MLX5_TXOFF_CONFIG_MULTI (1u << 0) /* Multi-segment packets.*/ -#define MLX5_TXOFF_CONFIG_TSO (1u << 1) /* TCP send offload supported.*/ -#define MLX5_TXOFF_CONFIG_SWP (1u << 2) /* Tunnels/SW Parser offloads.*/ -#define MLX5_TXOFF_CONFIG_CSUM (1u << 3) /* Check Sums offloaded. */ -#define MLX5_TXOFF_CONFIG_INLINE (1u << 4) /* Data inlining supported. */ -#define MLX5_TXOFF_CONFIG_VLAN (1u << 5) /* VLAN insertion supported.*/ -#define MLX5_TXOFF_CONFIG_METADATA (1u << 6) /* Flow metadata. */ -#define MLX5_TXOFF_CONFIG_EMPW (1u << 8) /* Enhanced MPW supported.*/ -#define MLX5_TXOFF_CONFIG_MPW (1u << 9) /* Legacy MPW supported.*/ -#define MLX5_TXOFF_CONFIG_TXPP (1u << 10) /* Scheduling on timestamp.*/ - -/* The most common offloads groups. */ -#define MLX5_TXOFF_CONFIG_NONE 0 -#define MLX5_TXOFF_CONFIG_FULL (MLX5_TXOFF_CONFIG_MULTI | \ - MLX5_TXOFF_CONFIG_TSO | \ - MLX5_TXOFF_CONFIG_SWP | \ - MLX5_TXOFF_CONFIG_CSUM | \ - MLX5_TXOFF_CONFIG_INLINE | \ - MLX5_TXOFF_CONFIG_VLAN | \ - MLX5_TXOFF_CONFIG_METADATA) - -#define MLX5_TXOFF_CONFIG(mask) (olx & MLX5_TXOFF_CONFIG_##mask) - -#define MLX5_TXOFF_DECL(func, olx) \ -static uint16_t mlx5_tx_burst_##func(void *txq, \ - struct rte_mbuf **pkts, \ - uint16_t pkts_n) \ -{ \ - return mlx5_tx_burst_tmpl((struct mlx5_txq_data *)txq, \ - pkts, pkts_n, (olx)); \ -} - #define MLX5_TXOFF_INFO(func, olx) {mlx5_tx_burst_##func, olx}, /* static asserts */ @@ -139,7 +88,6 @@ enum mlx5_txcmp_code { uint8_t mlx5_swp_types_table[1 << 10] __rte_cache_aligned; uint64_t rte_net_mlx5_dynf_inline_mask; -#define PKT_TX_DYNF_NOINLINE rte_net_mlx5_dynf_inline_mask /** * Build a table to translate Rx completion flags to packet type. @@ -366,109 +314,6 @@ enum mlx5_txcmp_code { } } -/** - * Set Software Parser flags and offsets in Ethernet Segment of WQE. - * Flags must be preliminary initialized to zero. - * - * @param loc - * Pointer to burst routine local context. - * @param swp_flags - * Pointer to store Software Parser flags - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Software Parser offsets packed in dword. - * Software Parser flags are set by pointer. - */ -static __rte_always_inline uint32_t -txq_mbuf_to_swp(struct mlx5_txq_local *__rte_restrict loc, - uint8_t *swp_flags, - unsigned int olx) -{ - uint64_t ol, tunnel; - unsigned int idx, off; - uint32_t set; - - if (!MLX5_TXOFF_CONFIG(SWP)) - return 0; - ol = loc->mbuf->ol_flags; - tunnel = ol & PKT_TX_TUNNEL_MASK; - /* - * Check whether Software Parser is required. - * Only customized tunnels may ask for. - */ - if (likely(tunnel != PKT_TX_TUNNEL_UDP && tunnel != PKT_TX_TUNNEL_IP)) - return 0; - /* - * The index should have: - * bit[0:1] = PKT_TX_L4_MASK - * bit[4] = PKT_TX_IPV6 - * bit[8] = PKT_TX_OUTER_IPV6 - * bit[9] = PKT_TX_OUTER_UDP - */ - idx = (ol & (PKT_TX_L4_MASK | PKT_TX_IPV6 | PKT_TX_OUTER_IPV6)) >> 52; - idx |= (tunnel == PKT_TX_TUNNEL_UDP) ? (1 << 9) : 0; - *swp_flags = mlx5_swp_types_table[idx]; - /* - * Set offsets for SW parser. Since ConnectX-5, SW parser just - * complements HW parser. SW parser starts to engage only if HW parser - * can't reach a header. For the older devices, HW parser will not kick - * in if any of SWP offsets is set. Therefore, all of the L3 offsets - * should be set regardless of HW offload. - */ - off = loc->mbuf->outer_l2_len; - if (MLX5_TXOFF_CONFIG(VLAN) && ol & PKT_TX_VLAN_PKT) - off += sizeof(struct rte_vlan_hdr); - set = (off >> 1) << 8; /* Outer L3 offset. */ - off += loc->mbuf->outer_l3_len; - if (tunnel == PKT_TX_TUNNEL_UDP) - set |= off >> 1; /* Outer L4 offset. */ - if (ol & (PKT_TX_IPV4 | PKT_TX_IPV6)) { /* Inner IP. */ - const uint64_t csum = ol & PKT_TX_L4_MASK; - off += loc->mbuf->l2_len; - set |= (off >> 1) << 24; /* Inner L3 offset. */ - if (csum == PKT_TX_TCP_CKSUM || - csum == PKT_TX_UDP_CKSUM || - (MLX5_TXOFF_CONFIG(TSO) && ol & PKT_TX_TCP_SEG)) { - off += loc->mbuf->l3_len; - set |= (off >> 1) << 16; /* Inner L4 offset. */ - } - } - set = rte_cpu_to_le_32(set); - return set; -} - -/** - * Convert the Checksum offloads to Verbs. - * - * @param buf - * Pointer to the mbuf. - * - * @return - * Converted checksum flags. - */ -static __rte_always_inline uint8_t -txq_ol_cksum_to_cs(struct rte_mbuf *buf) -{ - uint32_t idx; - uint8_t is_tunnel = !!(buf->ol_flags & PKT_TX_TUNNEL_MASK); - const uint64_t ol_flags_mask = PKT_TX_TCP_SEG | PKT_TX_L4_MASK | - PKT_TX_IP_CKSUM | PKT_TX_OUTER_IP_CKSUM; - - /* - * The index should have: - * bit[0] = PKT_TX_TCP_SEG - * bit[2:3] = PKT_TX_UDP_CKSUM, PKT_TX_TCP_CKSUM - * bit[4] = PKT_TX_IP_CKSUM - * bit[8] = PKT_TX_OUTER_IP_CKSUM - * bit[9] = tunnel - */ - idx = ((buf->ol_flags & ol_flags_mask) >> 50) | (!!is_tunnel << 9); - return mlx5_cksum_table[idx]; -} - #define MLX5_SYSTEM_LOG_DIR "/var/log" /** * Dump debug information to log file. @@ -720,214 +565,6 @@ enum mlx5_txcmp_code { } /** - * Free the mbufs from the linear array of pointers. - * - * @param txq - * Pointer to Tx queue structure. - * @param pkts - * Pointer to array of packets to be free. - * @param pkts_n - * Number of packets to be freed. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_free_mbuf(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - unsigned int olx __rte_unused) -{ - struct rte_mempool *pool = NULL; - struct rte_mbuf **p_free = NULL; - struct rte_mbuf *mbuf; - unsigned int n_free = 0; - - /* - * The implemented algorithm eliminates - * copying pointers to temporary array - * for rte_mempool_put_bulk() calls. - */ - MLX5_ASSERT(pkts); - MLX5_ASSERT(pkts_n); - /* - * Free mbufs directly to the pool in bulk - * if fast free offload is engaged - */ - if (!MLX5_TXOFF_CONFIG(MULTI) && txq->fast_free) { - mbuf = *pkts; - pool = mbuf->pool; - rte_mempool_put_bulk(pool, (void *)pkts, pkts_n); - return; - } - for (;;) { - for (;;) { - /* - * Decrement mbuf reference counter, detach - * indirect and external buffers if needed. - */ - mbuf = rte_pktmbuf_prefree_seg(*pkts); - if (likely(mbuf != NULL)) { - MLX5_ASSERT(mbuf == *pkts); - if (likely(n_free != 0)) { - if (unlikely(pool != mbuf->pool)) - /* From different pool. */ - break; - } else { - /* Start new scan array. */ - pool = mbuf->pool; - p_free = pkts; - } - ++n_free; - ++pkts; - --pkts_n; - if (unlikely(pkts_n == 0)) { - mbuf = NULL; - break; - } - } else { - /* - * This happens if mbuf is still referenced. - * We can't put it back to the pool, skip. - */ - ++pkts; - --pkts_n; - if (unlikely(n_free != 0)) - /* There is some array to free.*/ - break; - if (unlikely(pkts_n == 0)) - /* Last mbuf, nothing to free. */ - return; - } - } - for (;;) { - /* - * This loop is implemented to avoid multiple - * inlining of rte_mempool_put_bulk(). - */ - MLX5_ASSERT(pool); - MLX5_ASSERT(p_free); - MLX5_ASSERT(n_free); - /* - * Free the array of pre-freed mbufs - * belonging to the same memory pool. - */ - rte_mempool_put_bulk(pool, (void *)p_free, n_free); - if (unlikely(mbuf != NULL)) { - /* There is the request to start new scan. */ - pool = mbuf->pool; - p_free = pkts++; - n_free = 1; - --pkts_n; - if (likely(pkts_n != 0)) - break; - /* - * This is the last mbuf to be freed. - * Do one more loop iteration to complete. - * This is rare case of the last unique mbuf. - */ - mbuf = NULL; - continue; - } - if (likely(pkts_n == 0)) - return; - n_free = 0; - break; - } - } -} -/* - * No inline version to free buffers for optimal call - * on the tx_burst completion. - */ -static __rte_noinline void -__mlx5_tx_free_mbuf(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - unsigned int olx __rte_unused) -{ - mlx5_tx_free_mbuf(txq, pkts, pkts_n, olx); -} - -/** - * Free the mbuf from the elts ring buffer till new tail. - * - * @param txq - * Pointer to Tx queue structure. - * @param tail - * Index in elts to free up to, becomes new elts tail. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_free_elts(struct mlx5_txq_data *__rte_restrict txq, - uint16_t tail, - unsigned int olx __rte_unused) -{ - uint16_t n_elts = tail - txq->elts_tail; - - MLX5_ASSERT(n_elts); - MLX5_ASSERT(n_elts <= txq->elts_s); - /* - * Implement a loop to support ring buffer wraparound - * with single inlining of mlx5_tx_free_mbuf(). - */ - do { - unsigned int part; - - part = txq->elts_s - (txq->elts_tail & txq->elts_m); - part = RTE_MIN(part, n_elts); - MLX5_ASSERT(part); - MLX5_ASSERT(part <= txq->elts_s); - mlx5_tx_free_mbuf(txq, - &txq->elts[txq->elts_tail & txq->elts_m], - part, olx); - txq->elts_tail += part; - n_elts -= part; - } while (n_elts); -} - -/** - * Store the mbuf being sent into elts ring buffer. - * On Tx completion these mbufs will be freed. - * - * @param txq - * Pointer to Tx queue structure. - * @param pkts - * Pointer to array of packets to be stored. - * @param pkts_n - * Number of packets to be stored. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_copy_elts(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - unsigned int olx __rte_unused) -{ - unsigned int part; - struct rte_mbuf **elts = (struct rte_mbuf **)txq->elts; - - MLX5_ASSERT(pkts); - MLX5_ASSERT(pkts_n); - part = txq->elts_s - (txq->elts_head & txq->elts_m); - MLX5_ASSERT(part); - MLX5_ASSERT(part <= txq->elts_s); - /* This code is a good candidate for vectorizing with SIMD. */ - rte_memcpy((void *)(elts + (txq->elts_head & txq->elts_m)), - (void *)pkts, - RTE_MIN(part, pkts_n) * sizeof(struct rte_mbuf *)); - txq->elts_head += pkts_n; - if (unlikely(part < pkts_n)) - /* The copy is wrapping around the elts array. */ - rte_memcpy((void *)elts, (void *)(pkts + part), - (pkts_n - part) * sizeof(struct rte_mbuf *)); -} - -/** * Update completion queue consuming index via doorbell * and flush the completed data buffers. * @@ -970,7 +607,7 @@ enum mlx5_txcmp_code { * NOTE: not inlined intentionally, it makes tx_burst * routine smaller, simple and faster - from experiments. */ -static void +void mlx5_tx_handle_completion(struct mlx5_txq_data *__rte_restrict txq, unsigned int olx __rte_unused) { @@ -1047,54 +684,6 @@ enum mlx5_txcmp_code { } /** - * Check if the completion request flag should be set in the last WQE. - * Both pushed mbufs and WQEs are monitored and the completion request - * flag is set if any of thresholds is reached. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_request_completion(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - uint16_t head = txq->elts_head; - unsigned int part; - - part = MLX5_TXOFF_CONFIG(INLINE) ? - 0 : loc->pkts_sent - loc->pkts_copy; - head += part; - if ((uint16_t)(head - txq->elts_comp) >= MLX5_TX_COMP_THRESH || - (MLX5_TXOFF_CONFIG(INLINE) && - (uint16_t)(txq->wqe_ci - txq->wqe_comp) >= txq->wqe_thres)) { - volatile struct mlx5_wqe *last = loc->wqe_last; - - MLX5_ASSERT(last); - txq->elts_comp = head; - if (MLX5_TXOFF_CONFIG(INLINE)) - txq->wqe_comp = txq->wqe_ci; - /* Request unconditional completion on last WQE. */ - last->cseg.flags = RTE_BE32(MLX5_COMP_ALWAYS << - MLX5_COMP_MODE_OFFSET); - /* Save elts_head in dedicated free on completion queue. */ -#ifdef RTE_LIBRTE_MLX5_DEBUG - txq->fcqs[txq->cq_pi++ & txq->cqe_m] = head | - (last->cseg.opcode >> 8) << 16; -#else - txq->fcqs[txq->cq_pi++ & txq->cqe_m] = head; -#endif - /* A CQE slot must always be available. */ - MLX5_ASSERT((txq->cq_pi - txq->cq_ci) <= txq->cqe_s); - } -} - -/** * DPDK callback to check the status of a tx descriptor. * * @param tx_queue @@ -1118,2924 +707,6 @@ enum mlx5_txcmp_code { return RTE_ETH_TX_DESC_DONE; } -/** - * Build the Control Segment with specified opcode: - * - MLX5_OPCODE_SEND - * - MLX5_OPCODE_ENHANCED_MPSW - * - MLX5_OPCODE_TSO - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Control Segment. - * @param ds - * Supposed length of WQE in segments. - * @param opcode - * SQ WQE opcode to put into Control Segment. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_cseg_init(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc __rte_unused, - struct mlx5_wqe *__rte_restrict wqe, - unsigned int ds, - unsigned int opcode, - unsigned int olx __rte_unused) -{ - struct mlx5_wqe_cseg *__rte_restrict cs = &wqe->cseg; - - /* For legacy MPW replace the EMPW by TSO with modifier. */ - if (MLX5_TXOFF_CONFIG(MPW) && opcode == MLX5_OPCODE_ENHANCED_MPSW) - opcode = MLX5_OPCODE_TSO | MLX5_OPC_MOD_MPW << 24; - cs->opcode = rte_cpu_to_be_32((txq->wqe_ci << 8) | opcode); - cs->sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); - cs->flags = RTE_BE32(MLX5_COMP_ONLY_FIRST_ERR << - MLX5_COMP_MODE_OFFSET); - cs->misc = RTE_BE32(0); -} - -/** - * Build the Synchronize Queue Segment with specified completion index. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Control Segment. - * @param wci - * Completion index in Clock Queue to wait. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_wseg_init(struct mlx5_txq_data *restrict txq, - struct mlx5_txq_local *restrict loc __rte_unused, - struct mlx5_wqe *restrict wqe, - unsigned int wci, - unsigned int olx __rte_unused) -{ - struct mlx5_wqe_qseg *qs; - - qs = RTE_PTR_ADD(wqe, MLX5_WSEG_SIZE); - qs->max_index = rte_cpu_to_be_32(wci); - qs->qpn_cqn = rte_cpu_to_be_32(txq->sh->txpp.clock_queue.cq_obj.cq->id); - qs->reserved0 = RTE_BE32(0); - qs->reserved1 = RTE_BE32(0); -} - -/** - * Build the Ethernet Segment without inlined data. - * Supports Software Parser, Checksums and VLAN - * insertion Tx offload features. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Ethernet Segment. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_eseg_none(struct mlx5_txq_data *__rte_restrict txq __rte_unused, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe *__rte_restrict wqe, - unsigned int olx) -{ - struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; - uint32_t csum; - - /* - * Calculate and set check sum flags first, dword field - * in segment may be shared with Software Parser flags. - */ - csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; - es->flags = rte_cpu_to_le_32(csum); - /* - * Calculate and set Software Parser offsets and flags. - * These flags a set for custom UDP and IP tunnel packets. - */ - es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); - /* Fill metadata field if needed. */ - es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? - loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? - *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; - /* Engage VLAN tag insertion feature if requested. */ - if (MLX5_TXOFF_CONFIG(VLAN) && - loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { - /* - * We should get here only if device support - * this feature correctly. - */ - MLX5_ASSERT(txq->vlan_en); - es->inline_hdr = rte_cpu_to_be_32(MLX5_ETH_WQE_VLAN_INSERT | - loc->mbuf->vlan_tci); - } else { - es->inline_hdr = RTE_BE32(0); - } -} - -/** - * Build the Ethernet Segment with minimal inlined data - * of MLX5_ESEG_MIN_INLINE_SIZE bytes length. This is - * used to fill the gap in single WQEBB WQEs. - * Supports Software Parser, Checksums and VLAN - * insertion Tx offload features. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Ethernet Segment. - * @param vlan - * Length of VLAN tag insertion if any. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_eseg_dmin(struct mlx5_txq_data *__rte_restrict txq __rte_unused, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe *__rte_restrict wqe, - unsigned int vlan, - unsigned int olx) -{ - struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; - uint32_t csum; - uint8_t *psrc, *pdst; - - /* - * Calculate and set check sum flags first, dword field - * in segment may be shared with Software Parser flags. - */ - csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; - es->flags = rte_cpu_to_le_32(csum); - /* - * Calculate and set Software Parser offsets and flags. - * These flags a set for custom UDP and IP tunnel packets. - */ - es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); - /* Fill metadata field if needed. */ - es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? - loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? - *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; - psrc = rte_pktmbuf_mtod(loc->mbuf, uint8_t *); - es->inline_hdr_sz = RTE_BE16(MLX5_ESEG_MIN_INLINE_SIZE); - es->inline_data = *(unaligned_uint16_t *)psrc; - psrc += sizeof(uint16_t); - pdst = (uint8_t *)(es + 1); - if (MLX5_TXOFF_CONFIG(VLAN) && vlan) { - /* Implement VLAN tag insertion as part inline data. */ - memcpy(pdst, psrc, 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t)); - pdst += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); - psrc += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); - /* Insert VLAN ethertype + VLAN tag. */ - *(unaligned_uint32_t *)pdst = rte_cpu_to_be_32 - ((RTE_ETHER_TYPE_VLAN << 16) | - loc->mbuf->vlan_tci); - pdst += sizeof(struct rte_vlan_hdr); - /* Copy the rest two bytes from packet data. */ - MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, sizeof(uint16_t))); - *(uint16_t *)pdst = *(unaligned_uint16_t *)psrc; - } else { - /* Fill the gap in the title WQEBB with inline data. */ - rte_mov16(pdst, psrc); - } -} - -/** - * Build the Ethernet Segment with entire packet - * data inlining. Checks the boundary of WQEBB and - * ring buffer wrapping, supports Software Parser, - * Checksums and VLAN insertion Tx offload features. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Ethernet Segment. - * @param vlan - * Length of VLAN tag insertion if any. - * @param inlen - * Length of data to inline (VLAN included, if any). - * @param tso - * TSO flag, set mss field from the packet. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Pointer to the next Data Segment (aligned and wrapped around). - */ -static __rte_always_inline struct mlx5_wqe_dseg * -mlx5_tx_eseg_data(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe *__rte_restrict wqe, - unsigned int vlan, - unsigned int inlen, - unsigned int tso, - unsigned int olx) -{ - struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; - uint32_t csum; - uint8_t *psrc, *pdst; - unsigned int part; - - /* - * Calculate and set check sum flags first, dword field - * in segment may be shared with Software Parser flags. - */ - csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; - if (tso) { - csum <<= 24; - csum |= loc->mbuf->tso_segsz; - es->flags = rte_cpu_to_be_32(csum); - } else { - es->flags = rte_cpu_to_le_32(csum); - } - /* - * Calculate and set Software Parser offsets and flags. - * These flags a set for custom UDP and IP tunnel packets. - */ - es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); - /* Fill metadata field if needed. */ - es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? - loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? - *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; - psrc = rte_pktmbuf_mtod(loc->mbuf, uint8_t *); - es->inline_hdr_sz = rte_cpu_to_be_16(inlen); - es->inline_data = *(unaligned_uint16_t *)psrc; - psrc += sizeof(uint16_t); - pdst = (uint8_t *)(es + 1); - if (MLX5_TXOFF_CONFIG(VLAN) && vlan) { - /* Implement VLAN tag insertion as part inline data. */ - memcpy(pdst, psrc, 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t)); - pdst += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); - psrc += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); - /* Insert VLAN ethertype + VLAN tag. */ - *(unaligned_uint32_t *)pdst = rte_cpu_to_be_32 - ((RTE_ETHER_TYPE_VLAN << 16) | - loc->mbuf->vlan_tci); - pdst += sizeof(struct rte_vlan_hdr); - /* Copy the rest two bytes from packet data. */ - MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, sizeof(uint16_t))); - *(uint16_t *)pdst = *(unaligned_uint16_t *)psrc; - psrc += sizeof(uint16_t); - } else { - /* Fill the gap in the title WQEBB with inline data. */ - rte_mov16(pdst, psrc); - psrc += sizeof(rte_v128u32_t); - } - pdst = (uint8_t *)(es + 2); - MLX5_ASSERT(inlen >= MLX5_ESEG_MIN_INLINE_SIZE); - MLX5_ASSERT(pdst < (uint8_t *)txq->wqes_end); - inlen -= MLX5_ESEG_MIN_INLINE_SIZE; - if (!inlen) { - MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE)); - return (struct mlx5_wqe_dseg *)pdst; - } - /* - * The WQEBB space availability is checked by caller. - * Here we should be aware of WQE ring buffer wraparound only. - */ - part = (uint8_t *)txq->wqes_end - pdst; - part = RTE_MIN(part, inlen); - do { - rte_memcpy(pdst, psrc, part); - inlen -= part; - if (likely(!inlen)) { - /* - * If return value is not used by the caller - * the code below will be optimized out. - */ - pdst += part; - pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); - if (unlikely(pdst >= (uint8_t *)txq->wqes_end)) - pdst = (uint8_t *)txq->wqes; - return (struct mlx5_wqe_dseg *)pdst; - } - pdst = (uint8_t *)txq->wqes; - psrc += part; - part = inlen; - } while (true); -} - -/** - * Copy data from chain of mbuf to the specified linear buffer. - * Checksums and VLAN insertion Tx offload features. If data - * from some mbuf copied completely this mbuf is freed. Local - * structure is used to keep the byte stream state. - * - * @param pdst - * Pointer to the destination linear buffer. - * @param loc - * Pointer to burst routine local context. - * @param len - * Length of data to be copied. - * @param must - * Length of data to be copied ignoring no inline hint. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Number of actual copied data bytes. This is always greater than or - * equal to must parameter and might be lesser than len in no inline - * hint flag is encountered. - */ -static __rte_always_inline unsigned int -mlx5_tx_mseg_memcpy(uint8_t *pdst, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int len, - unsigned int must, - unsigned int olx __rte_unused) -{ - struct rte_mbuf *mbuf; - unsigned int part, dlen, copy = 0; - uint8_t *psrc; - - MLX5_ASSERT(len); - MLX5_ASSERT(must <= len); - do { - /* Allow zero length packets, must check first. */ - dlen = rte_pktmbuf_data_len(loc->mbuf); - if (dlen <= loc->mbuf_off) { - /* Exhausted packet, just free. */ - mbuf = loc->mbuf; - loc->mbuf = mbuf->next; - rte_pktmbuf_free_seg(mbuf); - loc->mbuf_off = 0; - MLX5_ASSERT(loc->mbuf_nseg > 1); - MLX5_ASSERT(loc->mbuf); - --loc->mbuf_nseg; - if (loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE) { - unsigned int diff; - - if (copy >= must) { - /* - * We already copied the minimal - * requested amount of data. - */ - return copy; - } - diff = must - copy; - if (diff <= rte_pktmbuf_data_len(loc->mbuf)) { - /* - * Copy only the minimal required - * part of the data buffer. - */ - len = diff; - } - } - continue; - } - dlen -= loc->mbuf_off; - psrc = rte_pktmbuf_mtod_offset(loc->mbuf, uint8_t *, - loc->mbuf_off); - part = RTE_MIN(len, dlen); - rte_memcpy(pdst, psrc, part); - copy += part; - loc->mbuf_off += part; - len -= part; - if (!len) { - if (loc->mbuf_off >= rte_pktmbuf_data_len(loc->mbuf)) { - loc->mbuf_off = 0; - /* Exhausted packet, just free. */ - mbuf = loc->mbuf; - loc->mbuf = mbuf->next; - rte_pktmbuf_free_seg(mbuf); - loc->mbuf_off = 0; - MLX5_ASSERT(loc->mbuf_nseg >= 1); - --loc->mbuf_nseg; - } - return copy; - } - pdst += part; - } while (true); -} - -/** - * Build the Ethernet Segment with inlined data from - * multi-segment packet. Checks the boundary of WQEBB - * and ring buffer wrapping, supports Software Parser, - * Checksums and VLAN insertion Tx offload features. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Ethernet Segment. - * @param vlan - * Length of VLAN tag insertion if any. - * @param inlen - * Length of data to inline (VLAN included, if any). - * @param tso - * TSO flag, set mss field from the packet. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Pointer to the next Data Segment (aligned and - * possible NOT wrapped around - caller should do - * wrapping check on its own). - */ -static __rte_always_inline struct mlx5_wqe_dseg * -mlx5_tx_eseg_mdat(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe *__rte_restrict wqe, - unsigned int vlan, - unsigned int inlen, - unsigned int tso, - unsigned int olx) -{ - struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; - uint32_t csum; - uint8_t *pdst; - unsigned int part, tlen = 0; - - /* - * Calculate and set check sum flags first, uint32_t field - * in segment may be shared with Software Parser flags. - */ - csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; - if (tso) { - csum <<= 24; - csum |= loc->mbuf->tso_segsz; - es->flags = rte_cpu_to_be_32(csum); - } else { - es->flags = rte_cpu_to_le_32(csum); - } - /* - * Calculate and set Software Parser offsets and flags. - * These flags a set for custom UDP and IP tunnel packets. - */ - es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); - /* Fill metadata field if needed. */ - es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? - loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? - *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; - MLX5_ASSERT(inlen >= MLX5_ESEG_MIN_INLINE_SIZE); - pdst = (uint8_t *)&es->inline_data; - if (MLX5_TXOFF_CONFIG(VLAN) && vlan) { - /* Implement VLAN tag insertion as part inline data. */ - mlx5_tx_mseg_memcpy(pdst, loc, - 2 * RTE_ETHER_ADDR_LEN, - 2 * RTE_ETHER_ADDR_LEN, olx); - pdst += 2 * RTE_ETHER_ADDR_LEN; - *(unaligned_uint32_t *)pdst = rte_cpu_to_be_32 - ((RTE_ETHER_TYPE_VLAN << 16) | - loc->mbuf->vlan_tci); - pdst += sizeof(struct rte_vlan_hdr); - tlen += 2 * RTE_ETHER_ADDR_LEN + sizeof(struct rte_vlan_hdr); - } - MLX5_ASSERT(pdst < (uint8_t *)txq->wqes_end); - /* - * The WQEBB space availability is checked by caller. - * Here we should be aware of WQE ring buffer wraparound only. - */ - part = (uint8_t *)txq->wqes_end - pdst; - part = RTE_MIN(part, inlen - tlen); - MLX5_ASSERT(part); - do { - unsigned int copy; - - /* - * Copying may be interrupted inside the routine - * if run into no inline hint flag. - */ - copy = tlen >= txq->inlen_mode ? 0 : (txq->inlen_mode - tlen); - copy = mlx5_tx_mseg_memcpy(pdst, loc, part, copy, olx); - tlen += copy; - if (likely(inlen <= tlen) || copy < part) { - es->inline_hdr_sz = rte_cpu_to_be_16(tlen); - pdst += copy; - pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); - return (struct mlx5_wqe_dseg *)pdst; - } - pdst = (uint8_t *)txq->wqes; - part = inlen - tlen; - } while (true); -} - -/** - * Build the Data Segment of pointer type. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param dseg - * Pointer to WQE to fill with built Data Segment. - * @param buf - * Data buffer to point. - * @param len - * Data buffer length. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_dseg_ptr(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe_dseg *__rte_restrict dseg, - uint8_t *buf, - unsigned int len, - unsigned int olx __rte_unused) - -{ - MLX5_ASSERT(len); - dseg->bcount = rte_cpu_to_be_32(len); - dseg->lkey = mlx5_tx_mb2mr(txq, loc->mbuf); - dseg->pbuf = rte_cpu_to_be_64((uintptr_t)buf); -} - -/** - * Build the Data Segment of pointer type or inline - * if data length is less than buffer in minimal - * Data Segment size. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param dseg - * Pointer to WQE to fill with built Data Segment. - * @param buf - * Data buffer to point. - * @param len - * Data buffer length. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_dseg_iptr(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe_dseg *__rte_restrict dseg, - uint8_t *buf, - unsigned int len, - unsigned int olx __rte_unused) - -{ - uintptr_t dst, src; - - MLX5_ASSERT(len); - if (len > MLX5_DSEG_MIN_INLINE_SIZE) { - dseg->bcount = rte_cpu_to_be_32(len); - dseg->lkey = mlx5_tx_mb2mr(txq, loc->mbuf); - dseg->pbuf = rte_cpu_to_be_64((uintptr_t)buf); - - return; - } - dseg->bcount = rte_cpu_to_be_32(len | MLX5_ETH_WQE_DATA_INLINE); - /* Unrolled implementation of generic rte_memcpy. */ - dst = (uintptr_t)&dseg->inline_data[0]; - src = (uintptr_t)buf; - if (len & 0x08) { -#ifdef RTE_ARCH_STRICT_ALIGN - MLX5_ASSERT(dst == RTE_PTR_ALIGN(dst, sizeof(uint32_t))); - *(uint32_t *)dst = *(unaligned_uint32_t *)src; - dst += sizeof(uint32_t); - src += sizeof(uint32_t); - *(uint32_t *)dst = *(unaligned_uint32_t *)src; - dst += sizeof(uint32_t); - src += sizeof(uint32_t); -#else - *(uint64_t *)dst = *(unaligned_uint64_t *)src; - dst += sizeof(uint64_t); - src += sizeof(uint64_t); -#endif - } - if (len & 0x04) { - *(uint32_t *)dst = *(unaligned_uint32_t *)src; - dst += sizeof(uint32_t); - src += sizeof(uint32_t); - } - if (len & 0x02) { - *(uint16_t *)dst = *(unaligned_uint16_t *)src; - dst += sizeof(uint16_t); - src += sizeof(uint16_t); - } - if (len & 0x01) - *(uint8_t *)dst = *(uint8_t *)src; -} - -/** - * Build the Data Segment of inlined data from single - * segment packet, no VLAN insertion. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param dseg - * Pointer to WQE to fill with built Data Segment. - * @param buf - * Data buffer to point. - * @param len - * Data buffer length. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Pointer to the next Data Segment after inlined data. - * Ring buffer wraparound check is needed. We do not - * do it here because it may not be needed for the - * last packet in the eMPW session. - */ -static __rte_always_inline struct mlx5_wqe_dseg * -mlx5_tx_dseg_empw(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc __rte_unused, - struct mlx5_wqe_dseg *__rte_restrict dseg, - uint8_t *buf, - unsigned int len, - unsigned int olx __rte_unused) -{ - unsigned int part; - uint8_t *pdst; - - if (!MLX5_TXOFF_CONFIG(MPW)) { - /* Store the descriptor byte counter for eMPW sessions. */ - dseg->bcount = rte_cpu_to_be_32(len | MLX5_ETH_WQE_DATA_INLINE); - pdst = &dseg->inline_data[0]; - } else { - /* The entire legacy MPW session counter is stored on close. */ - pdst = (uint8_t *)dseg; - } - /* - * The WQEBB space availability is checked by caller. - * Here we should be aware of WQE ring buffer wraparound only. - */ - part = (uint8_t *)txq->wqes_end - pdst; - part = RTE_MIN(part, len); - do { - rte_memcpy(pdst, buf, part); - len -= part; - if (likely(!len)) { - pdst += part; - if (!MLX5_TXOFF_CONFIG(MPW)) - pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); - /* Note: no final wraparound check here. */ - return (struct mlx5_wqe_dseg *)pdst; - } - pdst = (uint8_t *)txq->wqes; - buf += part; - part = len; - } while (true); -} - -/** - * Build the Data Segment of inlined data from single - * segment packet with VLAN insertion. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param dseg - * Pointer to the dseg fill with built Data Segment. - * @param buf - * Data buffer to point. - * @param len - * Data buffer length. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Pointer to the next Data Segment after inlined data. - * Ring buffer wraparound check is needed. - */ -static __rte_always_inline struct mlx5_wqe_dseg * -mlx5_tx_dseg_vlan(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc __rte_unused, - struct mlx5_wqe_dseg *__rte_restrict dseg, - uint8_t *buf, - unsigned int len, - unsigned int olx __rte_unused) - -{ - unsigned int part; - uint8_t *pdst; - - MLX5_ASSERT(len > MLX5_ESEG_MIN_INLINE_SIZE); - if (!MLX5_TXOFF_CONFIG(MPW)) { - /* Store the descriptor byte counter for eMPW sessions. */ - dseg->bcount = rte_cpu_to_be_32 - ((len + sizeof(struct rte_vlan_hdr)) | - MLX5_ETH_WQE_DATA_INLINE); - pdst = &dseg->inline_data[0]; - } else { - /* The entire legacy MPW session counter is stored on close. */ - pdst = (uint8_t *)dseg; - } - memcpy(pdst, buf, MLX5_DSEG_MIN_INLINE_SIZE); - buf += MLX5_DSEG_MIN_INLINE_SIZE; - pdst += MLX5_DSEG_MIN_INLINE_SIZE; - len -= MLX5_DSEG_MIN_INLINE_SIZE; - /* Insert VLAN ethertype + VLAN tag. Pointer is aligned. */ - MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE)); - if (unlikely(pdst >= (uint8_t *)txq->wqes_end)) - pdst = (uint8_t *)txq->wqes; - *(uint32_t *)pdst = rte_cpu_to_be_32((RTE_ETHER_TYPE_VLAN << 16) | - loc->mbuf->vlan_tci); - pdst += sizeof(struct rte_vlan_hdr); - /* - * The WQEBB space availability is checked by caller. - * Here we should be aware of WQE ring buffer wraparound only. - */ - part = (uint8_t *)txq->wqes_end - pdst; - part = RTE_MIN(part, len); - do { - rte_memcpy(pdst, buf, part); - len -= part; - if (likely(!len)) { - pdst += part; - if (!MLX5_TXOFF_CONFIG(MPW)) - pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); - /* Note: no final wraparound check here. */ - return (struct mlx5_wqe_dseg *)pdst; - } - pdst = (uint8_t *)txq->wqes; - buf += part; - part = len; - } while (true); -} - -/** - * Build the Ethernet Segment with optionally inlined data with - * VLAN insertion and following Data Segments (if any) from - * multi-segment packet. Used by ordinary send and TSO. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param wqe - * Pointer to WQE to fill with built Ethernet/Data Segments. - * @param vlan - * Length of VLAN header to insert, 0 means no VLAN insertion. - * @param inlen - * Data length to inline. For TSO this parameter specifies - * exact value, for ordinary send routine can be aligned by - * caller to provide better WQE space saving and data buffer - * start address alignment. This length includes VLAN header - * being inserted. - * @param tso - * Zero means ordinary send, inlined data can be extended, - * otherwise this is TSO, inlined data length is fixed. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * Actual size of built WQE in segments. - */ -static __rte_always_inline unsigned int -mlx5_tx_mseg_build(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - struct mlx5_wqe *__rte_restrict wqe, - unsigned int vlan, - unsigned int inlen, - unsigned int tso, - unsigned int olx __rte_unused) -{ - struct mlx5_wqe_dseg *__rte_restrict dseg; - unsigned int ds; - - MLX5_ASSERT((rte_pktmbuf_pkt_len(loc->mbuf) + vlan) >= inlen); - loc->mbuf_nseg = NB_SEGS(loc->mbuf); - loc->mbuf_off = 0; - - dseg = mlx5_tx_eseg_mdat(txq, loc, wqe, vlan, inlen, tso, olx); - if (!loc->mbuf_nseg) - goto dseg_done; - /* - * There are still some mbuf remaining, not inlined. - * The first mbuf may be partially inlined and we - * must process the possible non-zero data offset. - */ - if (loc->mbuf_off) { - unsigned int dlen; - uint8_t *dptr; - - /* - * Exhausted packets must be dropped before. - * Non-zero offset means there are some data - * remained in the packet. - */ - MLX5_ASSERT(loc->mbuf_off < rte_pktmbuf_data_len(loc->mbuf)); - MLX5_ASSERT(rte_pktmbuf_data_len(loc->mbuf)); - dptr = rte_pktmbuf_mtod_offset(loc->mbuf, uint8_t *, - loc->mbuf_off); - dlen = rte_pktmbuf_data_len(loc->mbuf) - loc->mbuf_off; - /* - * Build the pointer/minimal data Data Segment. - * Do ring buffer wrapping check in advance. - */ - if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) - dseg = (struct mlx5_wqe_dseg *)txq->wqes; - mlx5_tx_dseg_iptr(txq, loc, dseg, dptr, dlen, olx); - /* Store the mbuf to be freed on completion. */ - MLX5_ASSERT(loc->elts_free); - txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; - --loc->elts_free; - ++dseg; - if (--loc->mbuf_nseg == 0) - goto dseg_done; - loc->mbuf = loc->mbuf->next; - loc->mbuf_off = 0; - } - do { - if (unlikely(!rte_pktmbuf_data_len(loc->mbuf))) { - struct rte_mbuf *mbuf; - - /* Zero length segment found, just skip. */ - mbuf = loc->mbuf; - loc->mbuf = loc->mbuf->next; - rte_pktmbuf_free_seg(mbuf); - if (--loc->mbuf_nseg == 0) - break; - } else { - if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) - dseg = (struct mlx5_wqe_dseg *)txq->wqes; - mlx5_tx_dseg_iptr - (txq, loc, dseg, - rte_pktmbuf_mtod(loc->mbuf, uint8_t *), - rte_pktmbuf_data_len(loc->mbuf), olx); - MLX5_ASSERT(loc->elts_free); - txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; - --loc->elts_free; - ++dseg; - if (--loc->mbuf_nseg == 0) - break; - loc->mbuf = loc->mbuf->next; - } - } while (true); - -dseg_done: - /* Calculate actual segments used from the dseg pointer. */ - if ((uintptr_t)wqe < (uintptr_t)dseg) - ds = ((uintptr_t)dseg - (uintptr_t)wqe) / MLX5_WSEG_SIZE; - else - ds = (((uintptr_t)dseg - (uintptr_t)wqe) + - txq->wqe_s * MLX5_WQE_SIZE) / MLX5_WSEG_SIZE; - return ds; -} - -/** - * The routine checks timestamp flag in the current packet, - * and push WAIT WQE into the queue if scheduling is required. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_SINGLE - continue processing with the packet. - * MLX5_TXCMP_CODE_MULTI - the WAIT inserted, continue processing. - * Local context variables partially updated. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_schedule_send(struct mlx5_txq_data *restrict txq, - struct mlx5_txq_local *restrict loc, - unsigned int olx) -{ - if (MLX5_TXOFF_CONFIG(TXPP) && - loc->mbuf->ol_flags & txq->ts_mask) { - struct mlx5_wqe *wqe; - uint64_t ts; - int32_t wci; - - /* - * Estimate the required space quickly and roughly. - * We would like to ensure the packet can be pushed - * to the queue and we won't get the orphan WAIT WQE. - */ - if (loc->wqe_free <= MLX5_WQE_SIZE_MAX / MLX5_WQE_SIZE || - loc->elts_free < NB_SEGS(loc->mbuf)) - return MLX5_TXCMP_CODE_EXIT; - /* Convert the timestamp into completion to wait. */ - ts = *RTE_MBUF_DYNFIELD(loc->mbuf, txq->ts_offset, uint64_t *); - wci = mlx5_txpp_convert_tx_ts(txq->sh, ts); - if (unlikely(wci < 0)) - return MLX5_TXCMP_CODE_SINGLE; - /* Build the WAIT WQE with specified completion. */ - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - mlx5_tx_cseg_init(txq, loc, wqe, 2, MLX5_OPCODE_WAIT, olx); - mlx5_tx_wseg_init(txq, loc, wqe, wci, olx); - ++txq->wqe_ci; - --loc->wqe_free; - return MLX5_TXCMP_CODE_MULTI; - } - return MLX5_TXCMP_CODE_SINGLE; -} - -/** - * Tx one packet function for multi-segment TSO. Supports all - * types of Tx offloads, uses MLX5_OPCODE_TSO to build WQEs, - * sends one packet per WQE. - * - * This routine is responsible for storing processed mbuf - * into elts ring buffer and update elts_head. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. - * Local context variables partially updated. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_packet_multi_tso(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - struct mlx5_wqe *__rte_restrict wqe; - unsigned int ds, dlen, inlen, ntcp, vlan = 0; - - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - /* - * Calculate data length to be inlined to estimate - * the required space in WQE ring buffer. - */ - dlen = rte_pktmbuf_pkt_len(loc->mbuf); - if (MLX5_TXOFF_CONFIG(VLAN) && loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) - vlan = sizeof(struct rte_vlan_hdr); - inlen = loc->mbuf->l2_len + vlan + - loc->mbuf->l3_len + loc->mbuf->l4_len; - if (unlikely((!inlen || !loc->mbuf->tso_segsz))) - return MLX5_TXCMP_CODE_ERROR; - if (loc->mbuf->ol_flags & PKT_TX_TUNNEL_MASK) - inlen += loc->mbuf->outer_l2_len + loc->mbuf->outer_l3_len; - /* Packet must contain all TSO headers. */ - if (unlikely(inlen > MLX5_MAX_TSO_HEADER || - inlen <= MLX5_ESEG_MIN_INLINE_SIZE || - inlen > (dlen + vlan))) - return MLX5_TXCMP_CODE_ERROR; - MLX5_ASSERT(inlen >= txq->inlen_mode); - /* - * Check whether there are enough free WQEBBs: - * - Control Segment - * - Ethernet Segment - * - First Segment of inlined Ethernet data - * - ... data continued ... - * - Data Segments of pointer/min inline type - */ - ds = NB_SEGS(loc->mbuf) + 2 + (inlen - - MLX5_ESEG_MIN_INLINE_SIZE + - MLX5_WSEG_SIZE + - MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; - if (unlikely(loc->wqe_free < ((ds + 3) / 4))) - return MLX5_TXCMP_CODE_EXIT; - /* Check for maximal WQE size. */ - if (unlikely((MLX5_WQE_SIZE_MAX / MLX5_WSEG_SIZE) < ((ds + 3) / 4))) - return MLX5_TXCMP_CODE_ERROR; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes/packets counters. */ - ntcp = (dlen - (inlen - vlan) + loc->mbuf->tso_segsz - 1) / - loc->mbuf->tso_segsz; - /* - * One will be added for mbuf itself - * at the end of the mlx5_tx_burst from - * loc->pkts_sent field. - */ - --ntcp; - txq->stats.opackets += ntcp; - txq->stats.obytes += dlen + vlan + ntcp * inlen; -#endif - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, 0, MLX5_OPCODE_TSO, olx); - ds = mlx5_tx_mseg_build(txq, loc, wqe, vlan, inlen, 1, olx); - wqe->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); - txq->wqe_ci += (ds + 3) / 4; - loc->wqe_free -= (ds + 3) / 4; - return MLX5_TXCMP_CODE_MULTI; -} - -/** - * Tx one packet function for multi-segment SEND. Supports all - * types of Tx offloads, uses MLX5_OPCODE_SEND to build WQEs, - * sends one packet per WQE, without any data inlining in - * Ethernet Segment. - * - * This routine is responsible for storing processed mbuf - * into elts ring buffer and update elts_head. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. - * Local context variables partially updated. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_packet_multi_send(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - struct mlx5_wqe_dseg *__rte_restrict dseg; - struct mlx5_wqe *__rte_restrict wqe; - unsigned int ds, nseg; - - MLX5_ASSERT(NB_SEGS(loc->mbuf) > 1); - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - /* - * No inline at all, it means the CPU cycles saving - * is prioritized at configuration, we should not - * copy any packet data to WQE. - */ - nseg = NB_SEGS(loc->mbuf); - ds = 2 + nseg; - if (unlikely(loc->wqe_free < ((ds + 3) / 4))) - return MLX5_TXCMP_CODE_EXIT; - /* Check for maximal WQE size. */ - if (unlikely((MLX5_WQE_SIZE_MAX / MLX5_WSEG_SIZE) < ((ds + 3) / 4))) - return MLX5_TXCMP_CODE_ERROR; - /* - * Some Tx offloads may cause an error if - * packet is not long enough, check against - * assumed minimal length. - */ - if (rte_pktmbuf_pkt_len(loc->mbuf) <= MLX5_ESEG_MIN_INLINE_SIZE) - return MLX5_TXCMP_CODE_ERROR; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - txq->stats.obytes += rte_pktmbuf_pkt_len(loc->mbuf); - if (MLX5_TXOFF_CONFIG(VLAN) && - loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) - txq->stats.obytes += sizeof(struct rte_vlan_hdr); -#endif - /* - * SEND WQE, one WQEBB: - * - Control Segment, SEND opcode - * - Ethernet Segment, optional VLAN, no inline - * - Data Segments, pointer only type - */ - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, ds, MLX5_OPCODE_SEND, olx); - mlx5_tx_eseg_none(txq, loc, wqe, olx); - dseg = &wqe->dseg[0]; - do { - if (unlikely(!rte_pktmbuf_data_len(loc->mbuf))) { - struct rte_mbuf *mbuf; - - /* - * Zero length segment found, have to - * correct total size of WQE in segments. - * It is supposed to be rare occasion, so - * in normal case (no zero length segments) - * we avoid extra writing to the Control - * Segment. - */ - --ds; - wqe->cseg.sq_ds -= RTE_BE32(1); - mbuf = loc->mbuf; - loc->mbuf = mbuf->next; - rte_pktmbuf_free_seg(mbuf); - if (--nseg == 0) - break; - } else { - mlx5_tx_dseg_ptr - (txq, loc, dseg, - rte_pktmbuf_mtod(loc->mbuf, uint8_t *), - rte_pktmbuf_data_len(loc->mbuf), olx); - txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; - --loc->elts_free; - if (--nseg == 0) - break; - ++dseg; - if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) - dseg = (struct mlx5_wqe_dseg *)txq->wqes; - loc->mbuf = loc->mbuf->next; - } - } while (true); - txq->wqe_ci += (ds + 3) / 4; - loc->wqe_free -= (ds + 3) / 4; - return MLX5_TXCMP_CODE_MULTI; -} - -/** - * Tx one packet function for multi-segment SEND. Supports all - * types of Tx offloads, uses MLX5_OPCODE_SEND to build WQEs, - * sends one packet per WQE, with data inlining in - * Ethernet Segment and minimal Data Segments. - * - * This routine is responsible for storing processed mbuf - * into elts ring buffer and update elts_head. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. - * Local context variables partially updated. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_packet_multi_inline(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - struct mlx5_wqe *__rte_restrict wqe; - unsigned int ds, inlen, dlen, vlan = 0; - - MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); - MLX5_ASSERT(NB_SEGS(loc->mbuf) > 1); - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - /* - * First calculate data length to be inlined - * to estimate the required space for WQE. - */ - dlen = rte_pktmbuf_pkt_len(loc->mbuf); - if (MLX5_TXOFF_CONFIG(VLAN) && loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) - vlan = sizeof(struct rte_vlan_hdr); - inlen = dlen + vlan; - /* Check against minimal length. */ - if (inlen <= MLX5_ESEG_MIN_INLINE_SIZE) - return MLX5_TXCMP_CODE_ERROR; - MLX5_ASSERT(txq->inlen_send >= MLX5_ESEG_MIN_INLINE_SIZE); - if (inlen > txq->inlen_send || - loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE) { - struct rte_mbuf *mbuf; - unsigned int nxlen; - uintptr_t start; - - /* - * Packet length exceeds the allowed inline - * data length, check whether the minimal - * inlining is required. - */ - if (txq->inlen_mode) { - MLX5_ASSERT(txq->inlen_mode >= - MLX5_ESEG_MIN_INLINE_SIZE); - MLX5_ASSERT(txq->inlen_mode <= txq->inlen_send); - inlen = txq->inlen_mode; - } else { - if (loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE || - !vlan || txq->vlan_en) { - /* - * VLAN insertion will be done inside by HW. - * It is not utmost effective - VLAN flag is - * checked twice, but we should proceed the - * inlining length correctly and take into - * account the VLAN header being inserted. - */ - return mlx5_tx_packet_multi_send - (txq, loc, olx); - } - inlen = MLX5_ESEG_MIN_INLINE_SIZE; - } - /* - * Now we know the minimal amount of data is requested - * to inline. Check whether we should inline the buffers - * from the chain beginning to eliminate some mbufs. - */ - mbuf = loc->mbuf; - nxlen = rte_pktmbuf_data_len(mbuf); - if (unlikely(nxlen <= txq->inlen_send)) { - /* We can inline first mbuf at least. */ - if (nxlen < inlen) { - unsigned int smlen; - - /* Scan mbufs till inlen filled. */ - do { - smlen = nxlen; - mbuf = NEXT(mbuf); - MLX5_ASSERT(mbuf); - nxlen = rte_pktmbuf_data_len(mbuf); - nxlen += smlen; - } while (unlikely(nxlen < inlen)); - if (unlikely(nxlen > txq->inlen_send)) { - /* We cannot inline entire mbuf. */ - smlen = inlen - smlen; - start = rte_pktmbuf_mtod_offset - (mbuf, uintptr_t, smlen); - goto do_align; - } - } - do { - inlen = nxlen; - mbuf = NEXT(mbuf); - /* There should be not end of packet. */ - MLX5_ASSERT(mbuf); - nxlen = inlen + rte_pktmbuf_data_len(mbuf); - } while (unlikely(nxlen < txq->inlen_send)); - } - start = rte_pktmbuf_mtod(mbuf, uintptr_t); - /* - * Check whether we can do inline to align start - * address of data buffer to cacheline. - */ -do_align: - start = (~start + 1) & (RTE_CACHE_LINE_SIZE - 1); - if (unlikely(start)) { - start += inlen; - if (start <= txq->inlen_send) - inlen = start; - } - } - /* - * Check whether there are enough free WQEBBs: - * - Control Segment - * - Ethernet Segment - * - First Segment of inlined Ethernet data - * - ... data continued ... - * - Data Segments of pointer/min inline type - * - * Estimate the number of Data Segments conservatively, - * supposing no any mbufs is being freed during inlining. - */ - MLX5_ASSERT(inlen <= txq->inlen_send); - ds = NB_SEGS(loc->mbuf) + 2 + (inlen - - MLX5_ESEG_MIN_INLINE_SIZE + - MLX5_WSEG_SIZE + - MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; - if (unlikely(loc->wqe_free < ((ds + 3) / 4))) - return MLX5_TXCMP_CODE_EXIT; - /* Check for maximal WQE size. */ - if (unlikely((MLX5_WQE_SIZE_MAX / MLX5_WSEG_SIZE) < ((ds + 3) / 4))) - return MLX5_TXCMP_CODE_ERROR; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes/packets counters. */ - txq->stats.obytes += dlen + vlan; -#endif - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, 0, MLX5_OPCODE_SEND, olx); - ds = mlx5_tx_mseg_build(txq, loc, wqe, vlan, inlen, 0, olx); - wqe->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); - txq->wqe_ci += (ds + 3) / 4; - loc->wqe_free -= (ds + 3) / 4; - return MLX5_TXCMP_CODE_MULTI; -} - -/** - * Tx burst function for multi-segment packets. Supports all - * types of Tx offloads, uses MLX5_OPCODE_SEND/TSO to build WQEs, - * sends one packet per WQE. Function stops sending if it - * encounters the single-segment packet. - * - * This routine is responsible for storing processed mbuf - * into elts ring buffer and update elts_head. - * - * @param txq - * Pointer to TX queue structure. - * @param[in] pkts - * Packets to transmit. - * @param pkts_n - * Number of packets in array. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. - * MLX5_TXCMP_CODE_SINGLE - single-segment packet encountered. - * MLX5_TXCMP_CODE_TSO - TSO single-segment packet encountered. - * Local context variables updated. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_burst_mseg(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - MLX5_ASSERT(loc->elts_free && loc->wqe_free); - MLX5_ASSERT(pkts_n > loc->pkts_sent); - pkts += loc->pkts_sent + 1; - pkts_n -= loc->pkts_sent; - for (;;) { - enum mlx5_txcmp_code ret; - - MLX5_ASSERT(NB_SEGS(loc->mbuf) > 1); - /* - * Estimate the number of free elts quickly but - * conservatively. Some segment may be fully inlined - * and freed, ignore this here - precise estimation - * is costly. - */ - if (loc->elts_free < NB_SEGS(loc->mbuf)) - return MLX5_TXCMP_CODE_EXIT; - if (MLX5_TXOFF_CONFIG(TSO) && - unlikely(loc->mbuf->ol_flags & PKT_TX_TCP_SEG)) { - /* Proceed with multi-segment TSO. */ - ret = mlx5_tx_packet_multi_tso(txq, loc, olx); - } else if (MLX5_TXOFF_CONFIG(INLINE)) { - /* Proceed with multi-segment SEND with inlining. */ - ret = mlx5_tx_packet_multi_inline(txq, loc, olx); - } else { - /* Proceed with multi-segment SEND w/o inlining. */ - ret = mlx5_tx_packet_multi_send(txq, loc, olx); - } - if (ret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (ret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - /* WQE is built, go to the next packet. */ - ++loc->pkts_sent; - --pkts_n; - if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - loc->mbuf = *pkts++; - if (pkts_n > 1) - rte_prefetch0(*pkts); - if (likely(NB_SEGS(loc->mbuf) > 1)) - continue; - /* Here ends the series of multi-segment packets. */ - if (MLX5_TXOFF_CONFIG(TSO) && - unlikely(loc->mbuf->ol_flags & PKT_TX_TCP_SEG)) - return MLX5_TXCMP_CODE_TSO; - return MLX5_TXCMP_CODE_SINGLE; - } - MLX5_ASSERT(false); -} - -/** - * Tx burst function for single-segment packets with TSO. - * Supports all types of Tx offloads, except multi-packets. - * Uses MLX5_OPCODE_TSO to build WQEs, sends one packet per WQE. - * Function stops sending if it encounters the multi-segment - * packet or packet without TSO requested. - * - * The routine is responsible for storing processed mbuf - * into elts ring buffer and update elts_head if inline - * offloads is requested due to possible early freeing - * of the inlined mbufs (can not store pkts array in elts - * as a batch). - * - * @param txq - * Pointer to TX queue structure. - * @param[in] pkts - * Packets to transmit. - * @param pkts_n - * Number of packets in array. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. - * MLX5_TXCMP_CODE_SINGLE - single-segment packet encountered. - * MLX5_TXCMP_CODE_MULTI - multi-segment packet encountered. - * Local context variables updated. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_burst_tso(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - MLX5_ASSERT(loc->elts_free && loc->wqe_free); - MLX5_ASSERT(pkts_n > loc->pkts_sent); - pkts += loc->pkts_sent + 1; - pkts_n -= loc->pkts_sent; - for (;;) { - struct mlx5_wqe_dseg *__rte_restrict dseg; - struct mlx5_wqe *__rte_restrict wqe; - unsigned int ds, dlen, hlen, ntcp, vlan = 0; - uint8_t *dptr; - - MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - dlen = rte_pktmbuf_data_len(loc->mbuf); - if (MLX5_TXOFF_CONFIG(VLAN) && - loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { - vlan = sizeof(struct rte_vlan_hdr); - } - /* - * First calculate the WQE size to check - * whether we have enough space in ring buffer. - */ - hlen = loc->mbuf->l2_len + vlan + - loc->mbuf->l3_len + loc->mbuf->l4_len; - if (unlikely((!hlen || !loc->mbuf->tso_segsz))) - return MLX5_TXCMP_CODE_ERROR; - if (loc->mbuf->ol_flags & PKT_TX_TUNNEL_MASK) - hlen += loc->mbuf->outer_l2_len + - loc->mbuf->outer_l3_len; - /* Segment must contain all TSO headers. */ - if (unlikely(hlen > MLX5_MAX_TSO_HEADER || - hlen <= MLX5_ESEG_MIN_INLINE_SIZE || - hlen > (dlen + vlan))) - return MLX5_TXCMP_CODE_ERROR; - /* - * Check whether there are enough free WQEBBs: - * - Control Segment - * - Ethernet Segment - * - First Segment of inlined Ethernet data - * - ... data continued ... - * - Finishing Data Segment of pointer type - */ - ds = 4 + (hlen - MLX5_ESEG_MIN_INLINE_SIZE + - MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; - if (loc->wqe_free < ((ds + 3) / 4)) - return MLX5_TXCMP_CODE_EXIT; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes/packets counters. */ - ntcp = (dlen + vlan - hlen + - loc->mbuf->tso_segsz - 1) / - loc->mbuf->tso_segsz; - /* - * One will be added for mbuf itself at the end - * of the mlx5_tx_burst from loc->pkts_sent field. - */ - --ntcp; - txq->stats.opackets += ntcp; - txq->stats.obytes += dlen + vlan + ntcp * hlen; -#endif - /* - * Build the TSO WQE: - * - Control Segment - * - Ethernet Segment with hlen bytes inlined - * - Data Segment of pointer type - */ - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, ds, - MLX5_OPCODE_TSO, olx); - dseg = mlx5_tx_eseg_data(txq, loc, wqe, vlan, hlen, 1, olx); - dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *) + hlen - vlan; - dlen -= hlen - vlan; - mlx5_tx_dseg_ptr(txq, loc, dseg, dptr, dlen, olx); - /* - * WQE is built, update the loop parameters - * and go to the next packet. - */ - txq->wqe_ci += (ds + 3) / 4; - loc->wqe_free -= (ds + 3) / 4; - if (MLX5_TXOFF_CONFIG(INLINE)) - txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; - --loc->elts_free; - ++loc->pkts_sent; - --pkts_n; - if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - loc->mbuf = *pkts++; - if (pkts_n > 1) - rte_prefetch0(*pkts); - if (MLX5_TXOFF_CONFIG(MULTI) && - unlikely(NB_SEGS(loc->mbuf) > 1)) - return MLX5_TXCMP_CODE_MULTI; - if (likely(!(loc->mbuf->ol_flags & PKT_TX_TCP_SEG))) - return MLX5_TXCMP_CODE_SINGLE; - /* Continue with the next TSO packet. */ - } - MLX5_ASSERT(false); -} - -/** - * Analyze the packet and select the best method to send. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * @param newp - * The predefined flag whether do complete check for - * multi-segment packets and TSO. - * - * @return - * MLX5_TXCMP_CODE_MULTI - multi-segment packet encountered. - * MLX5_TXCMP_CODE_TSO - TSO required, use TSO/LSO. - * MLX5_TXCMP_CODE_SINGLE - single-segment packet, use SEND. - * MLX5_TXCMP_CODE_EMPW - single-segment packet, use MPW. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_able_to_empw(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx, - bool newp) -{ - /* Check for multi-segment packet. */ - if (newp && - MLX5_TXOFF_CONFIG(MULTI) && - unlikely(NB_SEGS(loc->mbuf) > 1)) - return MLX5_TXCMP_CODE_MULTI; - /* Check for TSO packet. */ - if (newp && - MLX5_TXOFF_CONFIG(TSO) && - unlikely(loc->mbuf->ol_flags & PKT_TX_TCP_SEG)) - return MLX5_TXCMP_CODE_TSO; - /* Check if eMPW is enabled at all. */ - if (!MLX5_TXOFF_CONFIG(EMPW)) - return MLX5_TXCMP_CODE_SINGLE; - /* Check if eMPW can be engaged. */ - if (MLX5_TXOFF_CONFIG(VLAN) && - unlikely(loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) && - (!MLX5_TXOFF_CONFIG(INLINE) || - unlikely((rte_pktmbuf_data_len(loc->mbuf) + - sizeof(struct rte_vlan_hdr)) > txq->inlen_empw))) { - /* - * eMPW does not support VLAN insertion offload, - * we have to inline the entire packet but - * packet is too long for inlining. - */ - return MLX5_TXCMP_CODE_SINGLE; - } - return MLX5_TXCMP_CODE_EMPW; -} - -/** - * Check the next packet attributes to match with the eMPW batch ones. - * In addition, for legacy MPW the packet length is checked either. - * - * @param txq - * Pointer to TX queue structure. - * @param es - * Pointer to Ethernet Segment of eMPW batch. - * @param loc - * Pointer to burst routine local context. - * @param dlen - * Length of previous packet in MPW descriptor. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * true - packet match with eMPW batch attributes. - * false - no match, eMPW should be restarted. - */ -static __rte_always_inline bool -mlx5_tx_match_empw(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_wqe_eseg *__rte_restrict es, - struct mlx5_txq_local *__rte_restrict loc, - uint32_t dlen, - unsigned int olx) -{ - uint8_t swp_flags = 0; - - /* Compare the checksum flags, if any. */ - if (MLX5_TXOFF_CONFIG(CSUM) && - txq_ol_cksum_to_cs(loc->mbuf) != es->cs_flags) - return false; - /* Compare the Software Parser offsets and flags. */ - if (MLX5_TXOFF_CONFIG(SWP) && - (es->swp_offs != txq_mbuf_to_swp(loc, &swp_flags, olx) || - es->swp_flags != swp_flags)) - return false; - /* Fill metadata field if needed. */ - if (MLX5_TXOFF_CONFIG(METADATA) && - es->metadata != (loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? - *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0)) - return false; - /* Legacy MPW can send packets with the same lengt only. */ - if (MLX5_TXOFF_CONFIG(MPW) && - dlen != rte_pktmbuf_data_len(loc->mbuf)) - return false; - /* There must be no VLAN packets in eMPW loop. */ - if (MLX5_TXOFF_CONFIG(VLAN)) - MLX5_ASSERT(!(loc->mbuf->ol_flags & PKT_TX_VLAN_PKT)); - /* Check if the scheduling is requested. */ - if (MLX5_TXOFF_CONFIG(TXPP) && - loc->mbuf->ol_flags & txq->ts_mask) - return false; - return true; -} - -/* - * Update send loop variables and WQE for eMPW loop - * without data inlining. Number of Data Segments is - * equal to the number of sent packets. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param ds - * Number of packets/Data Segments/Packets. - * @param slen - * Accumulated statistics, bytes sent - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * true - packet match with eMPW batch attributes. - * false - no match, eMPW should be restarted. - */ -static __rte_always_inline void -mlx5_tx_sdone_empw(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int ds, - unsigned int slen, - unsigned int olx __rte_unused) -{ - MLX5_ASSERT(!MLX5_TXOFF_CONFIG(INLINE)); -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - txq->stats.obytes += slen; -#else - (void)slen; -#endif - loc->elts_free -= ds; - loc->pkts_sent += ds; - ds += 2; - loc->wqe_last->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); - txq->wqe_ci += (ds + 3) / 4; - loc->wqe_free -= (ds + 3) / 4; -} - -/* - * Update send loop variables and WQE for eMPW loop - * with data inlining. Gets the size of pushed descriptors - * and data to the WQE. - * - * @param txq - * Pointer to TX queue structure. - * @param loc - * Pointer to burst routine local context. - * @param len - * Total size of descriptor/data in bytes. - * @param slen - * Accumulated statistics, data bytes sent. - * @param wqem - * The base WQE for the eMPW/MPW descriptor. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * true - packet match with eMPW batch attributes. - * false - no match, eMPW should be restarted. - */ -static __rte_always_inline void -mlx5_tx_idone_empw(struct mlx5_txq_data *__rte_restrict txq, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int len, - unsigned int slen, - struct mlx5_wqe *__rte_restrict wqem, - unsigned int olx __rte_unused) -{ - struct mlx5_wqe_dseg *dseg = &wqem->dseg[0]; - - MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - txq->stats.obytes += slen; -#else - (void)slen; -#endif - if (MLX5_TXOFF_CONFIG(MPW) && dseg->bcount == RTE_BE32(0)) { - /* - * If the legacy MPW session contains the inline packets - * we should set the only inline data segment length - * and align the total length to the segment size. - */ - MLX5_ASSERT(len > sizeof(dseg->bcount)); - dseg->bcount = rte_cpu_to_be_32((len - sizeof(dseg->bcount)) | - MLX5_ETH_WQE_DATA_INLINE); - len = (len + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE + 2; - } else { - /* - * The session is not legacy MPW or contains the - * data buffer pointer segments. - */ - MLX5_ASSERT((len % MLX5_WSEG_SIZE) == 0); - len = len / MLX5_WSEG_SIZE + 2; - } - wqem->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | len); - txq->wqe_ci += (len + 3) / 4; - loc->wqe_free -= (len + 3) / 4; - loc->wqe_last = wqem; -} - -/** - * The set of Tx burst functions for single-segment packets - * without TSO and with Multi-Packet Writing feature support. - * Supports all types of Tx offloads, except multi-packets - * and TSO. - * - * Uses MLX5_OPCODE_EMPW to build WQEs if possible and sends - * as many packet per WQE as it can. If eMPW is not configured - * or packet can not be sent with eMPW (VLAN insertion) the - * ordinary SEND opcode is used and only one packet placed - * in WQE. - * - * Functions stop sending if it encounters the multi-segment - * packet or packet with TSO requested. - * - * The routines are responsible for storing processed mbuf - * into elts ring buffer and update elts_head if inlining - * offload is requested. Otherwise the copying mbufs to elts - * can be postponed and completed at the end of burst routine. - * - * @param txq - * Pointer to TX queue structure. - * @param[in] pkts - * Packets to transmit. - * @param pkts_n - * Number of packets in array. - * @param loc - * Pointer to burst routine local context. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * @return - * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. - * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. - * MLX5_TXCMP_CODE_MULTI - multi-segment packet encountered. - * MLX5_TXCMP_CODE_TSO - TSO packet encountered. - * MLX5_TXCMP_CODE_SINGLE - used inside functions set. - * MLX5_TXCMP_CODE_EMPW - used inside functions set. - * - * Local context variables updated. - * - * - * The routine sends packets with MLX5_OPCODE_EMPW - * without inlining, this is dedicated optimized branch. - * No VLAN insertion is supported. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_burst_empw_simple(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - /* - * Subroutine is the part of mlx5_tx_burst_single() - * and sends single-segment packet with eMPW opcode - * without data inlining. - */ - MLX5_ASSERT(!MLX5_TXOFF_CONFIG(INLINE)); - MLX5_ASSERT(MLX5_TXOFF_CONFIG(EMPW)); - MLX5_ASSERT(loc->elts_free && loc->wqe_free); - MLX5_ASSERT(pkts_n > loc->pkts_sent); - pkts += loc->pkts_sent + 1; - pkts_n -= loc->pkts_sent; - for (;;) { - struct mlx5_wqe_dseg *__rte_restrict dseg; - struct mlx5_wqe_eseg *__rte_restrict eseg; - enum mlx5_txcmp_code ret; - unsigned int part, loop; - unsigned int slen = 0; - -next_empw: - MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - part = RTE_MIN(pkts_n, MLX5_TXOFF_CONFIG(MPW) ? - MLX5_MPW_MAX_PACKETS : - MLX5_EMPW_MAX_PACKETS); - if (unlikely(loc->elts_free < part)) { - /* We have no enough elts to save all mbufs. */ - if (unlikely(loc->elts_free < MLX5_EMPW_MIN_PACKETS)) - return MLX5_TXCMP_CODE_EXIT; - /* But we still able to send at least minimal eMPW. */ - part = loc->elts_free; - } - /* Check whether we have enough WQEs */ - if (unlikely(loc->wqe_free < ((2 + part + 3) / 4))) { - if (unlikely(loc->wqe_free < - ((2 + MLX5_EMPW_MIN_PACKETS + 3) / 4))) - return MLX5_TXCMP_CODE_EXIT; - part = (loc->wqe_free * 4) - 2; - } - if (likely(part > 1)) - rte_prefetch0(*pkts); - loc->wqe_last = txq->wqes + (txq->wqe_ci & txq->wqe_m); - /* - * Build eMPW title WQEBB: - * - Control Segment, eMPW opcode - * - Ethernet Segment, no inline - */ - mlx5_tx_cseg_init(txq, loc, loc->wqe_last, part + 2, - MLX5_OPCODE_ENHANCED_MPSW, olx); - mlx5_tx_eseg_none(txq, loc, loc->wqe_last, - olx & ~MLX5_TXOFF_CONFIG_VLAN); - eseg = &loc->wqe_last->eseg; - dseg = &loc->wqe_last->dseg[0]; - loop = part; - /* Store the packet length for legacy MPW. */ - if (MLX5_TXOFF_CONFIG(MPW)) - eseg->mss = rte_cpu_to_be_16 - (rte_pktmbuf_data_len(loc->mbuf)); - for (;;) { - uint32_t dlen = rte_pktmbuf_data_len(loc->mbuf); -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - slen += dlen; -#endif - mlx5_tx_dseg_ptr - (txq, loc, dseg, - rte_pktmbuf_mtod(loc->mbuf, uint8_t *), - dlen, olx); - if (unlikely(--loop == 0)) - break; - loc->mbuf = *pkts++; - if (likely(loop > 1)) - rte_prefetch0(*pkts); - ret = mlx5_tx_able_to_empw(txq, loc, olx, true); - /* - * Unroll the completion code to avoid - * returning variable value - it results in - * unoptimized sequent checking in caller. - */ - if (ret == MLX5_TXCMP_CODE_MULTI) { - part -= loop; - mlx5_tx_sdone_empw(txq, loc, part, slen, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - return MLX5_TXCMP_CODE_MULTI; - } - MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); - if (ret == MLX5_TXCMP_CODE_TSO) { - part -= loop; - mlx5_tx_sdone_empw(txq, loc, part, slen, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - return MLX5_TXCMP_CODE_TSO; - } - if (ret == MLX5_TXCMP_CODE_SINGLE) { - part -= loop; - mlx5_tx_sdone_empw(txq, loc, part, slen, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - return MLX5_TXCMP_CODE_SINGLE; - } - if (ret != MLX5_TXCMP_CODE_EMPW) { - MLX5_ASSERT(false); - part -= loop; - mlx5_tx_sdone_empw(txq, loc, part, slen, olx); - return MLX5_TXCMP_CODE_ERROR; - } - /* - * Check whether packet parameters coincide - * within assumed eMPW batch: - * - check sum settings - * - metadata value - * - software parser settings - * - packets length (legacy MPW only) - * - scheduling is not required - */ - if (!mlx5_tx_match_empw(txq, eseg, loc, dlen, olx)) { - MLX5_ASSERT(loop); - part -= loop; - mlx5_tx_sdone_empw(txq, loc, part, slen, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - pkts_n -= part; - goto next_empw; - } - /* Packet attributes match, continue the same eMPW. */ - ++dseg; - if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) - dseg = (struct mlx5_wqe_dseg *)txq->wqes; - } - /* eMPW is built successfully, update loop parameters. */ - MLX5_ASSERT(!loop); - MLX5_ASSERT(pkts_n >= part); -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - txq->stats.obytes += slen; -#endif - loc->elts_free -= part; - loc->pkts_sent += part; - txq->wqe_ci += (2 + part + 3) / 4; - loc->wqe_free -= (2 + part + 3) / 4; - pkts_n -= part; - if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - loc->mbuf = *pkts++; - ret = mlx5_tx_able_to_empw(txq, loc, olx, true); - if (unlikely(ret != MLX5_TXCMP_CODE_EMPW)) - return ret; - /* Continue sending eMPW batches. */ - } - MLX5_ASSERT(false); -} - -/** - * The routine sends packets with MLX5_OPCODE_EMPW - * with inlining, optionally supports VLAN insertion. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_burst_empw_inline(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - /* - * Subroutine is the part of mlx5_tx_burst_single() - * and sends single-segment packet with eMPW opcode - * with data inlining. - */ - MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); - MLX5_ASSERT(MLX5_TXOFF_CONFIG(EMPW)); - MLX5_ASSERT(loc->elts_free && loc->wqe_free); - MLX5_ASSERT(pkts_n > loc->pkts_sent); - pkts += loc->pkts_sent + 1; - pkts_n -= loc->pkts_sent; - for (;;) { - struct mlx5_wqe_dseg *__rte_restrict dseg; - struct mlx5_wqe *__rte_restrict wqem; - enum mlx5_txcmp_code ret; - unsigned int room, part, nlim; - unsigned int slen = 0; - - MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - /* - * Limits the amount of packets in one WQE - * to improve CQE latency generation. - */ - nlim = RTE_MIN(pkts_n, MLX5_TXOFF_CONFIG(MPW) ? - MLX5_MPW_INLINE_MAX_PACKETS : - MLX5_EMPW_MAX_PACKETS); - /* Check whether we have minimal amount WQEs */ - if (unlikely(loc->wqe_free < - ((2 + MLX5_EMPW_MIN_PACKETS + 3) / 4))) - return MLX5_TXCMP_CODE_EXIT; - if (likely(pkts_n > 1)) - rte_prefetch0(*pkts); - wqem = txq->wqes + (txq->wqe_ci & txq->wqe_m); - /* - * Build eMPW title WQEBB: - * - Control Segment, eMPW opcode, zero DS - * - Ethernet Segment, no inline - */ - mlx5_tx_cseg_init(txq, loc, wqem, 0, - MLX5_OPCODE_ENHANCED_MPSW, olx); - mlx5_tx_eseg_none(txq, loc, wqem, - olx & ~MLX5_TXOFF_CONFIG_VLAN); - dseg = &wqem->dseg[0]; - /* Store the packet length for legacy MPW. */ - if (MLX5_TXOFF_CONFIG(MPW)) - wqem->eseg.mss = rte_cpu_to_be_16 - (rte_pktmbuf_data_len(loc->mbuf)); - room = RTE_MIN(MLX5_WQE_SIZE_MAX / MLX5_WQE_SIZE, - loc->wqe_free) * MLX5_WQE_SIZE - - MLX5_WQE_CSEG_SIZE - - MLX5_WQE_ESEG_SIZE; - /* Limit the room for legacy MPW sessions for performance. */ - if (MLX5_TXOFF_CONFIG(MPW)) - room = RTE_MIN(room, - RTE_MAX(txq->inlen_empw + - sizeof(dseg->bcount) + - (MLX5_TXOFF_CONFIG(VLAN) ? - sizeof(struct rte_vlan_hdr) : 0), - MLX5_MPW_INLINE_MAX_PACKETS * - MLX5_WQE_DSEG_SIZE)); - /* Build WQE till we have space, packets and resources. */ - part = room; - for (;;) { - uint32_t dlen = rte_pktmbuf_data_len(loc->mbuf); - uint8_t *dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *); - unsigned int tlen; - - MLX5_ASSERT(room >= MLX5_WQE_DSEG_SIZE); - MLX5_ASSERT((room % MLX5_WQE_DSEG_SIZE) == 0); - MLX5_ASSERT((uintptr_t)dseg < (uintptr_t)txq->wqes_end); - /* - * Some Tx offloads may cause an error if - * packet is not long enough, check against - * assumed minimal length. - */ - if (unlikely(dlen <= MLX5_ESEG_MIN_INLINE_SIZE)) { - part -= room; - if (unlikely(!part)) - return MLX5_TXCMP_CODE_ERROR; - /* - * We have some successfully built - * packet Data Segments to send. - */ - mlx5_tx_idone_empw(txq, loc, part, - slen, wqem, olx); - return MLX5_TXCMP_CODE_ERROR; - } - /* Inline or not inline - that's the Question. */ - if (dlen > txq->inlen_empw || - loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE) - goto pointer_empw; - if (MLX5_TXOFF_CONFIG(MPW)) { - if (dlen > txq->inlen_send) - goto pointer_empw; - tlen = dlen; - if (part == room) { - /* Open new inline MPW session. */ - tlen += sizeof(dseg->bcount); - dseg->bcount = RTE_BE32(0); - dseg = RTE_PTR_ADD - (dseg, sizeof(dseg->bcount)); - } else { - /* - * No pointer and inline descriptor - * intermix for legacy MPW sessions. - */ - if (wqem->dseg[0].bcount) - break; - } - } else { - tlen = sizeof(dseg->bcount) + dlen; - } - /* Inline entire packet, optional VLAN insertion. */ - if (MLX5_TXOFF_CONFIG(VLAN) && - loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { - /* - * The packet length must be checked in - * mlx5_tx_able_to_empw() and packet - * fits into inline length guaranteed. - */ - MLX5_ASSERT((dlen + - sizeof(struct rte_vlan_hdr)) <= - txq->inlen_empw); - tlen += sizeof(struct rte_vlan_hdr); - if (room < tlen) - break; - dseg = mlx5_tx_dseg_vlan(txq, loc, dseg, - dptr, dlen, olx); -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - slen += sizeof(struct rte_vlan_hdr); -#endif - } else { - if (room < tlen) - break; - dseg = mlx5_tx_dseg_empw(txq, loc, dseg, - dptr, dlen, olx); - } - if (!MLX5_TXOFF_CONFIG(MPW)) - tlen = RTE_ALIGN(tlen, MLX5_WSEG_SIZE); - MLX5_ASSERT(room >= tlen); - room -= tlen; - /* - * Packet data are completely inline, - * we can try to free the packet. - */ - if (likely(loc->pkts_sent == loc->mbuf_free)) { - /* - * All the packets from the burst beginning - * are inline, we can free mbufs directly - * from the origin array on tx_burst exit(). - */ - loc->mbuf_free++; - goto next_mbuf; - } - /* - * In order no to call rte_pktmbuf_free_seg() here, - * in the most inner loop (that might be very - * expensive) we just save the mbuf in elts. - */ - txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; - loc->elts_free--; - goto next_mbuf; -pointer_empw: - /* - * No pointer and inline descriptor - * intermix for legacy MPW sessions. - */ - if (MLX5_TXOFF_CONFIG(MPW) && - part != room && - wqem->dseg[0].bcount == RTE_BE32(0)) - break; - /* - * Not inlinable VLAN packets are - * proceeded outside of this routine. - */ - MLX5_ASSERT(room >= MLX5_WQE_DSEG_SIZE); - if (MLX5_TXOFF_CONFIG(VLAN)) - MLX5_ASSERT(!(loc->mbuf->ol_flags & - PKT_TX_VLAN_PKT)); - mlx5_tx_dseg_ptr(txq, loc, dseg, dptr, dlen, olx); - /* We have to store mbuf in elts.*/ - txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; - loc->elts_free--; - room -= MLX5_WQE_DSEG_SIZE; - /* Ring buffer wraparound is checked at the loop end.*/ - ++dseg; -next_mbuf: -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - slen += dlen; -#endif - loc->pkts_sent++; - pkts_n--; - if (unlikely(!pkts_n || !loc->elts_free)) { - /* - * We have no resources/packets to - * continue build descriptors. - */ - part -= room; - mlx5_tx_idone_empw(txq, loc, part, - slen, wqem, olx); - return MLX5_TXCMP_CODE_EXIT; - } - loc->mbuf = *pkts++; - if (likely(pkts_n > 1)) - rte_prefetch0(*pkts); - ret = mlx5_tx_able_to_empw(txq, loc, olx, true); - /* - * Unroll the completion code to avoid - * returning variable value - it results in - * unoptimized sequent checking in caller. - */ - if (ret == MLX5_TXCMP_CODE_MULTI) { - part -= room; - mlx5_tx_idone_empw(txq, loc, part, - slen, wqem, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - return MLX5_TXCMP_CODE_MULTI; - } - MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); - if (ret == MLX5_TXCMP_CODE_TSO) { - part -= room; - mlx5_tx_idone_empw(txq, loc, part, - slen, wqem, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - return MLX5_TXCMP_CODE_TSO; - } - if (ret == MLX5_TXCMP_CODE_SINGLE) { - part -= room; - mlx5_tx_idone_empw(txq, loc, part, - slen, wqem, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - return MLX5_TXCMP_CODE_SINGLE; - } - if (ret != MLX5_TXCMP_CODE_EMPW) { - MLX5_ASSERT(false); - part -= room; - mlx5_tx_idone_empw(txq, loc, part, - slen, wqem, olx); - return MLX5_TXCMP_CODE_ERROR; - } - /* Check if we have minimal room left. */ - nlim--; - if (unlikely(!nlim || room < MLX5_WQE_DSEG_SIZE)) - break; - /* - * Check whether packet parameters coincide - * within assumed eMPW batch: - * - check sum settings - * - metadata value - * - software parser settings - * - packets length (legacy MPW only) - * - scheduling is not required - */ - if (!mlx5_tx_match_empw(txq, &wqem->eseg, - loc, dlen, olx)) - break; - /* Packet attributes match, continue the same eMPW. */ - if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) - dseg = (struct mlx5_wqe_dseg *)txq->wqes; - } - /* - * We get here to close an existing eMPW - * session and start the new one. - */ - MLX5_ASSERT(pkts_n); - part -= room; - if (unlikely(!part)) - return MLX5_TXCMP_CODE_EXIT; - mlx5_tx_idone_empw(txq, loc, part, slen, wqem, olx); - if (unlikely(!loc->elts_free || - !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - /* Continue the loop with new eMPW session. */ - } - MLX5_ASSERT(false); -} - -/** - * The routine sends packets with ordinary MLX5_OPCODE_SEND. - * Data inlining and VLAN insertion are supported. - */ -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_burst_single_send(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - /* - * Subroutine is the part of mlx5_tx_burst_single() - * and sends single-segment packet with SEND opcode. - */ - MLX5_ASSERT(loc->elts_free && loc->wqe_free); - MLX5_ASSERT(pkts_n > loc->pkts_sent); - pkts += loc->pkts_sent + 1; - pkts_n -= loc->pkts_sent; - for (;;) { - struct mlx5_wqe *__rte_restrict wqe; - enum mlx5_txcmp_code ret; - - MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); - if (MLX5_TXOFF_CONFIG(TXPP)) { - enum mlx5_txcmp_code wret; - - /* Generate WAIT for scheduling if requested. */ - wret = mlx5_tx_schedule_send(txq, loc, olx); - if (wret == MLX5_TXCMP_CODE_EXIT) - return MLX5_TXCMP_CODE_EXIT; - if (wret == MLX5_TXCMP_CODE_ERROR) - return MLX5_TXCMP_CODE_ERROR; - } - if (MLX5_TXOFF_CONFIG(INLINE)) { - unsigned int inlen, vlan = 0; - - inlen = rte_pktmbuf_data_len(loc->mbuf); - if (MLX5_TXOFF_CONFIG(VLAN) && - loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { - vlan = sizeof(struct rte_vlan_hdr); - inlen += vlan; - } - /* - * If inlining is enabled at configuration time - * the limit must be not less than minimal size. - * Otherwise we would do extra check for data - * size to avoid crashes due to length overflow. - */ - MLX5_ASSERT(txq->inlen_send >= - MLX5_ESEG_MIN_INLINE_SIZE); - if (inlen <= txq->inlen_send) { - unsigned int seg_n, wqe_n; - - rte_prefetch0(rte_pktmbuf_mtod - (loc->mbuf, uint8_t *)); - /* Check against minimal length. */ - if (inlen <= MLX5_ESEG_MIN_INLINE_SIZE) - return MLX5_TXCMP_CODE_ERROR; - if (loc->mbuf->ol_flags & - PKT_TX_DYNF_NOINLINE) { - /* - * The hint flag not to inline packet - * data is set. Check whether we can - * follow the hint. - */ - if ((!MLX5_TXOFF_CONFIG(EMPW) && - txq->inlen_mode) || - (MLX5_TXOFF_CONFIG(MPW) && - txq->inlen_mode)) { - if (inlen <= txq->inlen_send) - goto single_inline; - /* - * The hardware requires the - * minimal inline data header. - */ - goto single_min_inline; - } - if (MLX5_TXOFF_CONFIG(VLAN) && - vlan && !txq->vlan_en) { - /* - * We must insert VLAN tag - * by software means. - */ - goto single_part_inline; - } - goto single_no_inline; - } -single_inline: - /* - * Completely inlined packet data WQE: - * - Control Segment, SEND opcode - * - Ethernet Segment, no VLAN insertion - * - Data inlined, VLAN optionally inserted - * - Alignment to MLX5_WSEG_SIZE - * Have to estimate amount of WQEBBs - */ - seg_n = (inlen + 3 * MLX5_WSEG_SIZE - - MLX5_ESEG_MIN_INLINE_SIZE + - MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; - /* Check if there are enough WQEBBs. */ - wqe_n = (seg_n + 3) / 4; - if (wqe_n > loc->wqe_free) - return MLX5_TXCMP_CODE_EXIT; - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, seg_n, - MLX5_OPCODE_SEND, olx); - mlx5_tx_eseg_data(txq, loc, wqe, - vlan, inlen, 0, olx); - txq->wqe_ci += wqe_n; - loc->wqe_free -= wqe_n; - /* - * Packet data are completely inlined, - * free the packet immediately. - */ - rte_pktmbuf_free_seg(loc->mbuf); - } else if ((!MLX5_TXOFF_CONFIG(EMPW) || - MLX5_TXOFF_CONFIG(MPW)) && - txq->inlen_mode) { - /* - * If minimal inlining is requested the eMPW - * feature should be disabled due to data is - * inlined into Ethernet Segment, which can - * not contain inlined data for eMPW due to - * segment shared for all packets. - */ - struct mlx5_wqe_dseg *__rte_restrict dseg; - unsigned int ds; - uint8_t *dptr; - - /* - * The inline-mode settings require - * to inline the specified amount of - * data bytes to the Ethernet Segment. - * We should check the free space in - * WQE ring buffer to inline partially. - */ -single_min_inline: - MLX5_ASSERT(txq->inlen_send >= txq->inlen_mode); - MLX5_ASSERT(inlen > txq->inlen_mode); - MLX5_ASSERT(txq->inlen_mode >= - MLX5_ESEG_MIN_INLINE_SIZE); - /* - * Check whether there are enough free WQEBBs: - * - Control Segment - * - Ethernet Segment - * - First Segment of inlined Ethernet data - * - ... data continued ... - * - Finishing Data Segment of pointer type - */ - ds = (MLX5_WQE_CSEG_SIZE + - MLX5_WQE_ESEG_SIZE + - MLX5_WQE_DSEG_SIZE + - txq->inlen_mode - - MLX5_ESEG_MIN_INLINE_SIZE + - MLX5_WQE_DSEG_SIZE + - MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; - if (loc->wqe_free < ((ds + 3) / 4)) - return MLX5_TXCMP_CODE_EXIT; - /* - * Build the ordinary SEND WQE: - * - Control Segment - * - Ethernet Segment, inline inlen_mode bytes - * - Data Segment of pointer type - */ - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, ds, - MLX5_OPCODE_SEND, olx); - dseg = mlx5_tx_eseg_data(txq, loc, wqe, vlan, - txq->inlen_mode, - 0, olx); - dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *) + - txq->inlen_mode - vlan; - inlen -= txq->inlen_mode; - mlx5_tx_dseg_ptr(txq, loc, dseg, - dptr, inlen, olx); - /* - * WQE is built, update the loop parameters - * and got to the next packet. - */ - txq->wqe_ci += (ds + 3) / 4; - loc->wqe_free -= (ds + 3) / 4; - /* We have to store mbuf in elts.*/ - MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); - txq->elts[txq->elts_head++ & txq->elts_m] = - loc->mbuf; - --loc->elts_free; - } else { - uint8_t *dptr; - unsigned int dlen; - - /* - * Partially inlined packet data WQE, we have - * some space in title WQEBB, we can fill it - * with some packet data. It takes one WQEBB, - * it is available, no extra space check: - * - Control Segment, SEND opcode - * - Ethernet Segment, no VLAN insertion - * - MLX5_ESEG_MIN_INLINE_SIZE bytes of Data - * - Data Segment, pointer type - * - * We also get here if VLAN insertion is not - * supported by HW, the inline is enabled. - */ -single_part_inline: - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, 4, - MLX5_OPCODE_SEND, olx); - mlx5_tx_eseg_dmin(txq, loc, wqe, vlan, olx); - dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *) + - MLX5_ESEG_MIN_INLINE_SIZE - vlan; - /* - * The length check is performed above, by - * comparing with txq->inlen_send. We should - * not get overflow here. - */ - MLX5_ASSERT(inlen > MLX5_ESEG_MIN_INLINE_SIZE); - dlen = inlen - MLX5_ESEG_MIN_INLINE_SIZE; - mlx5_tx_dseg_ptr(txq, loc, &wqe->dseg[1], - dptr, dlen, olx); - ++txq->wqe_ci; - --loc->wqe_free; - /* We have to store mbuf in elts.*/ - MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); - txq->elts[txq->elts_head++ & txq->elts_m] = - loc->mbuf; - --loc->elts_free; - } -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - txq->stats.obytes += vlan + - rte_pktmbuf_data_len(loc->mbuf); -#endif - } else { - /* - * No inline at all, it means the CPU cycles saving - * is prioritized at configuration, we should not - * copy any packet data to WQE. - * - * SEND WQE, one WQEBB: - * - Control Segment, SEND opcode - * - Ethernet Segment, optional VLAN, no inline - * - Data Segment, pointer type - */ -single_no_inline: - wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); - loc->wqe_last = wqe; - mlx5_tx_cseg_init(txq, loc, wqe, 3, - MLX5_OPCODE_SEND, olx); - mlx5_tx_eseg_none(txq, loc, wqe, olx); - mlx5_tx_dseg_ptr - (txq, loc, &wqe->dseg[0], - rte_pktmbuf_mtod(loc->mbuf, uint8_t *), - rte_pktmbuf_data_len(loc->mbuf), olx); - ++txq->wqe_ci; - --loc->wqe_free; - /* - * We should not store mbuf pointer in elts - * if no inlining is configured, this is done - * by calling routine in a batch copy. - */ - MLX5_ASSERT(!MLX5_TXOFF_CONFIG(INLINE)); - --loc->elts_free; -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Update sent data bytes counter. */ - txq->stats.obytes += rte_pktmbuf_data_len(loc->mbuf); - if (MLX5_TXOFF_CONFIG(VLAN) && - loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) - txq->stats.obytes += - sizeof(struct rte_vlan_hdr); -#endif - } - ++loc->pkts_sent; - --pkts_n; - if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) - return MLX5_TXCMP_CODE_EXIT; - loc->mbuf = *pkts++; - if (pkts_n > 1) - rte_prefetch0(*pkts); - ret = mlx5_tx_able_to_empw(txq, loc, olx, true); - if (unlikely(ret != MLX5_TXCMP_CODE_SINGLE)) - return ret; - } - MLX5_ASSERT(false); -} - -static __rte_always_inline enum mlx5_txcmp_code -mlx5_tx_burst_single(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - unsigned int pkts_n, - struct mlx5_txq_local *__rte_restrict loc, - unsigned int olx) -{ - enum mlx5_txcmp_code ret; - - ret = mlx5_tx_able_to_empw(txq, loc, olx, false); - if (ret == MLX5_TXCMP_CODE_SINGLE) - goto ordinary_send; - MLX5_ASSERT(ret == MLX5_TXCMP_CODE_EMPW); - for (;;) { - /* Optimize for inline/no inline eMPW send. */ - ret = (MLX5_TXOFF_CONFIG(INLINE)) ? - mlx5_tx_burst_empw_inline - (txq, pkts, pkts_n, loc, olx) : - mlx5_tx_burst_empw_simple - (txq, pkts, pkts_n, loc, olx); - if (ret != MLX5_TXCMP_CODE_SINGLE) - return ret; - /* The resources to send one packet should remain. */ - MLX5_ASSERT(loc->elts_free && loc->wqe_free); -ordinary_send: - ret = mlx5_tx_burst_single_send(txq, pkts, pkts_n, loc, olx); - MLX5_ASSERT(ret != MLX5_TXCMP_CODE_SINGLE); - if (ret != MLX5_TXCMP_CODE_EMPW) - return ret; - /* The resources to send one packet should remain. */ - MLX5_ASSERT(loc->elts_free && loc->wqe_free); - } -} - -/** - * DPDK Tx callback template. This is configured template - * used to generate routines optimized for specified offload setup. - * One of this generated functions is chosen at SQ configuration - * time. - * - * @param txq - * Generic pointer to TX queue structure. - * @param[in] pkts - * Packets to transmit. - * @param pkts_n - * Number of packets in array. - * @param olx - * Configured offloads mask, presents the bits of MLX5_TXOFF_CONFIG_xxx - * values. Should be static to take compile time static configuration - * advantages. - * - * @return - * Number of packets successfully transmitted (<= pkts_n). - */ -static __rte_always_inline uint16_t -mlx5_tx_burst_tmpl(struct mlx5_txq_data *__rte_restrict txq, - struct rte_mbuf **__rte_restrict pkts, - uint16_t pkts_n, - unsigned int olx) -{ - struct mlx5_txq_local loc; - enum mlx5_txcmp_code ret; - unsigned int part; - - MLX5_ASSERT(txq->elts_s >= (uint16_t)(txq->elts_head - txq->elts_tail)); - MLX5_ASSERT(txq->wqe_s >= (uint16_t)(txq->wqe_ci - txq->wqe_pi)); - if (unlikely(!pkts_n)) - return 0; - if (MLX5_TXOFF_CONFIG(INLINE)) - loc.mbuf_free = 0; - loc.pkts_sent = 0; - loc.pkts_copy = 0; - loc.wqe_last = NULL; - -send_loop: - loc.pkts_loop = loc.pkts_sent; - /* - * Check if there are some CQEs, if any: - * - process an encountered errors - * - process the completed WQEs - * - free related mbufs - * - doorbell the NIC about processed CQEs - */ - rte_prefetch0(*(pkts + loc.pkts_sent)); - mlx5_tx_handle_completion(txq, olx); - /* - * Calculate the number of available resources - elts and WQEs. - * There are two possible different scenarios: - * - no data inlining into WQEs, one WQEBB may contains up to - * four packets, in this case elts become scarce resource - * - data inlining into WQEs, one packet may require multiple - * WQEBBs, the WQEs become the limiting factor. - */ - MLX5_ASSERT(txq->elts_s >= (uint16_t)(txq->elts_head - txq->elts_tail)); - loc.elts_free = txq->elts_s - - (uint16_t)(txq->elts_head - txq->elts_tail); - MLX5_ASSERT(txq->wqe_s >= (uint16_t)(txq->wqe_ci - txq->wqe_pi)); - loc.wqe_free = txq->wqe_s - - (uint16_t)(txq->wqe_ci - txq->wqe_pi); - if (unlikely(!loc.elts_free || !loc.wqe_free)) - goto burst_exit; - for (;;) { - /* - * Fetch the packet from array. Usually this is - * the first packet in series of multi/single - * segment packets. - */ - loc.mbuf = *(pkts + loc.pkts_sent); - /* Dedicated branch for multi-segment packets. */ - if (MLX5_TXOFF_CONFIG(MULTI) && - unlikely(NB_SEGS(loc.mbuf) > 1)) { - /* - * Multi-segment packet encountered. - * Hardware is able to process it only - * with SEND/TSO opcodes, one packet - * per WQE, do it in dedicated routine. - */ -enter_send_multi: - MLX5_ASSERT(loc.pkts_sent >= loc.pkts_copy); - part = loc.pkts_sent - loc.pkts_copy; - if (!MLX5_TXOFF_CONFIG(INLINE) && part) { - /* - * There are some single-segment mbufs not - * stored in elts. The mbufs must be in the - * same order as WQEs, so we must copy the - * mbufs to elts here, before the coming - * multi-segment packet mbufs is appended. - */ - mlx5_tx_copy_elts(txq, pkts + loc.pkts_copy, - part, olx); - loc.pkts_copy = loc.pkts_sent; - } - MLX5_ASSERT(pkts_n > loc.pkts_sent); - ret = mlx5_tx_burst_mseg(txq, pkts, pkts_n, &loc, olx); - if (!MLX5_TXOFF_CONFIG(INLINE)) - loc.pkts_copy = loc.pkts_sent; - /* - * These returned code checks are supposed - * to be optimized out due to routine inlining. - */ - if (ret == MLX5_TXCMP_CODE_EXIT) { - /* - * The routine returns this code when - * all packets are sent or there is no - * enough resources to complete request. - */ - break; - } - if (ret == MLX5_TXCMP_CODE_ERROR) { - /* - * The routine returns this code when - * some error in the incoming packets - * format occurred. - */ - txq->stats.oerrors++; - break; - } - if (ret == MLX5_TXCMP_CODE_SINGLE) { - /* - * The single-segment packet was encountered - * in the array, try to send it with the - * best optimized way, possible engaging eMPW. - */ - goto enter_send_single; - } - if (MLX5_TXOFF_CONFIG(TSO) && - ret == MLX5_TXCMP_CODE_TSO) { - /* - * The single-segment TSO packet was - * encountered in the array. - */ - goto enter_send_tso; - } - /* We must not get here. Something is going wrong. */ - MLX5_ASSERT(false); - txq->stats.oerrors++; - break; - } - /* Dedicated branch for single-segment TSO packets. */ - if (MLX5_TXOFF_CONFIG(TSO) && - unlikely(loc.mbuf->ol_flags & PKT_TX_TCP_SEG)) { - /* - * TSO might require special way for inlining - * (dedicated parameters) and is sent with - * MLX5_OPCODE_TSO opcode only, provide this - * in dedicated branch. - */ -enter_send_tso: - MLX5_ASSERT(NB_SEGS(loc.mbuf) == 1); - MLX5_ASSERT(pkts_n > loc.pkts_sent); - ret = mlx5_tx_burst_tso(txq, pkts, pkts_n, &loc, olx); - /* - * These returned code checks are supposed - * to be optimized out due to routine inlining. - */ - if (ret == MLX5_TXCMP_CODE_EXIT) - break; - if (ret == MLX5_TXCMP_CODE_ERROR) { - txq->stats.oerrors++; - break; - } - if (ret == MLX5_TXCMP_CODE_SINGLE) - goto enter_send_single; - if (MLX5_TXOFF_CONFIG(MULTI) && - ret == MLX5_TXCMP_CODE_MULTI) { - /* - * The multi-segment packet was - * encountered in the array. - */ - goto enter_send_multi; - } - /* We must not get here. Something is going wrong. */ - MLX5_ASSERT(false); - txq->stats.oerrors++; - break; - } - /* - * The dedicated branch for the single-segment packets - * without TSO. Often these ones can be sent using - * MLX5_OPCODE_EMPW with multiple packets in one WQE. - * The routine builds the WQEs till it encounters - * the TSO or multi-segment packet (in case if these - * offloads are requested at SQ configuration time). - */ -enter_send_single: - MLX5_ASSERT(pkts_n > loc.pkts_sent); - ret = mlx5_tx_burst_single(txq, pkts, pkts_n, &loc, olx); - /* - * These returned code checks are supposed - * to be optimized out due to routine inlining. - */ - if (ret == MLX5_TXCMP_CODE_EXIT) - break; - if (ret == MLX5_TXCMP_CODE_ERROR) { - txq->stats.oerrors++; - break; - } - if (MLX5_TXOFF_CONFIG(MULTI) && - ret == MLX5_TXCMP_CODE_MULTI) { - /* - * The multi-segment packet was - * encountered in the array. - */ - goto enter_send_multi; - } - if (MLX5_TXOFF_CONFIG(TSO) && - ret == MLX5_TXCMP_CODE_TSO) { - /* - * The single-segment TSO packet was - * encountered in the array. - */ - goto enter_send_tso; - } - /* We must not get here. Something is going wrong. */ - MLX5_ASSERT(false); - txq->stats.oerrors++; - break; - } - /* - * Main Tx loop is completed, do the rest: - * - set completion request if thresholds are reached - * - doorbell the hardware - * - copy the rest of mbufs to elts (if any) - */ - MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE) || - loc.pkts_sent >= loc.pkts_copy); - /* Take a shortcut if nothing is sent. */ - if (unlikely(loc.pkts_sent == loc.pkts_loop)) - goto burst_exit; - /* Request CQE generation if limits are reached. */ - mlx5_tx_request_completion(txq, &loc, olx); - /* - * Ring QP doorbell immediately after WQE building completion - * to improve latencies. The pure software related data treatment - * can be completed after doorbell. Tx CQEs for this SQ are - * processed in this thread only by the polling. - * - * The rdma core library can map doorbell register in two ways, - * depending on the environment variable "MLX5_SHUT_UP_BF": - * - * - as regular cached memory, the variable is either missing or - * set to zero. This type of mapping may cause the significant - * doorbell register writing latency and requires explicit - * memory write barrier to mitigate this issue and prevent - * write combining. - * - * - as non-cached memory, the variable is present and set to - * not "0" value. This type of mapping may cause performance - * impact under heavy loading conditions but the explicit write - * memory barrier is not required and it may improve core - * performance. - * - * - the legacy behaviour (prior 19.08 release) was to use some - * heuristics to decide whether write memory barrier should - * be performed. This behavior is supported with specifying - * tx_db_nc=2, write barrier is skipped if application - * provides the full recommended burst of packets, it - * supposes the next packets are coming and the write barrier - * will be issued on the next burst (after descriptor writing, - * at least). - */ - mlx5_tx_dbrec_cond_wmb(txq, loc.wqe_last, !txq->db_nc && - (!txq->db_heu || pkts_n % MLX5_TX_DEFAULT_BURST)); - /* Not all of the mbufs may be stored into elts yet. */ - part = MLX5_TXOFF_CONFIG(INLINE) ? 0 : loc.pkts_sent - loc.pkts_copy; - if (!MLX5_TXOFF_CONFIG(INLINE) && part) { - /* - * There are some single-segment mbufs not stored in elts. - * It can be only if the last packet was single-segment. - * The copying is gathered into one place due to it is - * a good opportunity to optimize that with SIMD. - * Unfortunately if inlining is enabled the gaps in - * pointer array may happen due to early freeing of the - * inlined mbufs. - */ - mlx5_tx_copy_elts(txq, pkts + loc.pkts_copy, part, olx); - loc.pkts_copy = loc.pkts_sent; - } - MLX5_ASSERT(txq->elts_s >= (uint16_t)(txq->elts_head - txq->elts_tail)); - MLX5_ASSERT(txq->wqe_s >= (uint16_t)(txq->wqe_ci - txq->wqe_pi)); - if (pkts_n > loc.pkts_sent) { - /* - * If burst size is large there might be no enough CQE - * fetched from completion queue and no enough resources - * freed to send all the packets. - */ - goto send_loop; - } -burst_exit: -#ifdef MLX5_PMD_SOFT_COUNTERS - /* Increment sent packets counter. */ - txq->stats.opackets += loc.pkts_sent; -#endif - if (MLX5_TXOFF_CONFIG(INLINE) && loc.mbuf_free) - __mlx5_tx_free_mbuf(txq, pkts, loc.mbuf_free, olx); - return loc.pkts_sent; -} - /* Generate routines with Enhanced Multi-Packet Write support. */ MLX5_TXOFF_DECL(full_empw, MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_EMPW) diff --git a/drivers/net/mlx5/mlx5_tx.h b/drivers/net/mlx5/mlx5_tx.h index 7f91d04..34843d4 100644 --- a/drivers/net/mlx5/mlx5_tx.h +++ b/drivers/net/mlx5/mlx5_tx.h @@ -20,8 +20,64 @@ #include "mlx5_autoconf.h" #include "mlx5_mr.h" +/* TX burst subroutines return codes. */ +enum mlx5_txcmp_code { + MLX5_TXCMP_CODE_EXIT = 0, + MLX5_TXCMP_CODE_ERROR, + MLX5_TXCMP_CODE_SINGLE, + MLX5_TXCMP_CODE_MULTI, + MLX5_TXCMP_CODE_TSO, + MLX5_TXCMP_CODE_EMPW, +}; + +/* + * These defines are used to configure Tx burst routine option set supported + * at compile time. The not specified options are optimized out due to if + * conditions can be explicitly calculated at compile time. + * The offloads with bigger runtime check (require more CPU cycles toskip) + * overhead should have the bigger index - this is needed to select the better + * matching routine function if no exact match and some offloads are not + * actually requested. + */ +#define MLX5_TXOFF_CONFIG_MULTI (1u << 0) /* Multi-segment packets.*/ +#define MLX5_TXOFF_CONFIG_TSO (1u << 1) /* TCP send offload supported.*/ +#define MLX5_TXOFF_CONFIG_SWP (1u << 2) /* Tunnels/SW Parser offloads.*/ +#define MLX5_TXOFF_CONFIG_CSUM (1u << 3) /* Check Sums offloaded. */ +#define MLX5_TXOFF_CONFIG_INLINE (1u << 4) /* Data inlining supported. */ +#define MLX5_TXOFF_CONFIG_VLAN (1u << 5) /* VLAN insertion supported.*/ +#define MLX5_TXOFF_CONFIG_METADATA (1u << 6) /* Flow metadata. */ +#define MLX5_TXOFF_CONFIG_EMPW (1u << 8) /* Enhanced MPW supported.*/ +#define MLX5_TXOFF_CONFIG_MPW (1u << 9) /* Legacy MPW supported.*/ +#define MLX5_TXOFF_CONFIG_TXPP (1u << 10) /* Scheduling on timestamp.*/ + +/* The most common offloads groups. */ +#define MLX5_TXOFF_CONFIG_NONE 0 +#define MLX5_TXOFF_CONFIG_FULL (MLX5_TXOFF_CONFIG_MULTI | \ + MLX5_TXOFF_CONFIG_TSO | \ + MLX5_TXOFF_CONFIG_SWP | \ + MLX5_TXOFF_CONFIG_CSUM | \ + MLX5_TXOFF_CONFIG_INLINE | \ + MLX5_TXOFF_CONFIG_VLAN | \ + MLX5_TXOFF_CONFIG_METADATA) + +#define MLX5_TXOFF_CONFIG(mask) (olx & MLX5_TXOFF_CONFIG_##mask) + +#define MLX5_TXOFF_DECL(func, olx) \ +static uint16_t mlx5_tx_burst_##func(void *txq, \ + struct rte_mbuf **pkts, \ + uint16_t pkts_n) \ +{ \ + return mlx5_tx_burst_tmpl((struct mlx5_txq_data *)txq, \ + pkts, pkts_n, (olx)); \ +} + /* Mbuf dynamic flag offset for inline. */ extern uint64_t rte_net_mlx5_dynf_inline_mask; +#define PKT_TX_DYNF_NOINLINE rte_net_mlx5_dynf_inline_mask + +extern uint32_t mlx5_ptype_table[] __rte_cache_aligned; +extern uint8_t mlx5_cksum_table[1 << 10] __rte_cache_aligned; +extern uint8_t mlx5_swp_types_table[1 << 10] __rte_cache_aligned; struct mlx5_txq_stats { #ifdef MLX5_PMD_SOFT_COUNTERS @@ -167,6 +223,8 @@ struct mlx5_txq_ctrl *mlx5_txq_hairpin_new uint16_t removed_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n); +void mlx5_tx_handle_completion(struct mlx5_txq_data *__rte_restrict txq, + unsigned int olx __rte_unused); int mlx5_tx_descriptor_status(void *tx_queue, uint16_t offset); void mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t queue_id, struct rte_eth_txq_info *qinfo); @@ -368,4 +426,3250 @@ uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, return ci; } +/** + * Set Software Parser flags and offsets in Ethernet Segment of WQE. + * Flags must be preliminary initialized to zero. + * + * @param loc + * Pointer to burst routine local context. + * @param swp_flags + * Pointer to store Software Parser flags. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Software Parser offsets packed in dword. + * Software Parser flags are set by pointer. + */ +static __rte_always_inline uint32_t +txq_mbuf_to_swp(struct mlx5_txq_local *__rte_restrict loc, + uint8_t *swp_flags, + unsigned int olx) +{ + uint64_t ol, tunnel; + unsigned int idx, off; + uint32_t set; + + if (!MLX5_TXOFF_CONFIG(SWP)) + return 0; + ol = loc->mbuf->ol_flags; + tunnel = ol & PKT_TX_TUNNEL_MASK; + /* + * Check whether Software Parser is required. + * Only customized tunnels may ask for. + */ + if (likely(tunnel != PKT_TX_TUNNEL_UDP && tunnel != PKT_TX_TUNNEL_IP)) + return 0; + /* + * The index should have: + * bit[0:1] = PKT_TX_L4_MASK + * bit[4] = PKT_TX_IPV6 + * bit[8] = PKT_TX_OUTER_IPV6 + * bit[9] = PKT_TX_OUTER_UDP + */ + idx = (ol & (PKT_TX_L4_MASK | PKT_TX_IPV6 | PKT_TX_OUTER_IPV6)) >> 52; + idx |= (tunnel == PKT_TX_TUNNEL_UDP) ? (1 << 9) : 0; + *swp_flags = mlx5_swp_types_table[idx]; + /* + * Set offsets for SW parser. Since ConnectX-5, SW parser just + * complements HW parser. SW parser starts to engage only if HW parser + * can't reach a header. For the older devices, HW parser will not kick + * in if any of SWP offsets is set. Therefore, all of the L3 offsets + * should be set regardless of HW offload. + */ + off = loc->mbuf->outer_l2_len; + if (MLX5_TXOFF_CONFIG(VLAN) && ol & PKT_TX_VLAN_PKT) + off += sizeof(struct rte_vlan_hdr); + set = (off >> 1) << 8; /* Outer L3 offset. */ + off += loc->mbuf->outer_l3_len; + if (tunnel == PKT_TX_TUNNEL_UDP) + set |= off >> 1; /* Outer L4 offset. */ + if (ol & (PKT_TX_IPV4 | PKT_TX_IPV6)) { /* Inner IP. */ + const uint64_t csum = ol & PKT_TX_L4_MASK; + off += loc->mbuf->l2_len; + set |= (off >> 1) << 24; /* Inner L3 offset. */ + if (csum == PKT_TX_TCP_CKSUM || + csum == PKT_TX_UDP_CKSUM || + (MLX5_TXOFF_CONFIG(TSO) && ol & PKT_TX_TCP_SEG)) { + off += loc->mbuf->l3_len; + set |= (off >> 1) << 16; /* Inner L4 offset. */ + } + } + set = rte_cpu_to_le_32(set); + return set; +} + +/** + * Convert the Checksum offloads to Verbs. + * + * @param buf + * Pointer to the mbuf. + * + * @return + * Converted checksum flags. + */ +static __rte_always_inline uint8_t +txq_ol_cksum_to_cs(struct rte_mbuf *buf) +{ + uint32_t idx; + uint8_t is_tunnel = !!(buf->ol_flags & PKT_TX_TUNNEL_MASK); + const uint64_t ol_flags_mask = PKT_TX_TCP_SEG | PKT_TX_L4_MASK | + PKT_TX_IP_CKSUM | PKT_TX_OUTER_IP_CKSUM; + + /* + * The index should have: + * bit[0] = PKT_TX_TCP_SEG + * bit[2:3] = PKT_TX_UDP_CKSUM, PKT_TX_TCP_CKSUM + * bit[4] = PKT_TX_IP_CKSUM + * bit[8] = PKT_TX_OUTER_IP_CKSUM + * bit[9] = tunnel + */ + idx = ((buf->ol_flags & ol_flags_mask) >> 50) | (!!is_tunnel << 9); + return mlx5_cksum_table[idx]; +} + +/** + * Free the mbufs from the linear array of pointers. + * + * @param txq + * Pointer to Tx queue structure. + * @param pkts + * Pointer to array of packets to be free. + * @param pkts_n + * Number of packets to be freed. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_free_mbuf(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + unsigned int olx __rte_unused) +{ + struct rte_mempool *pool = NULL; + struct rte_mbuf **p_free = NULL; + struct rte_mbuf *mbuf; + unsigned int n_free = 0; + + /* + * The implemented algorithm eliminates + * copying pointers to temporary array + * for rte_mempool_put_bulk() calls. + */ + MLX5_ASSERT(pkts); + MLX5_ASSERT(pkts_n); + /* + * Free mbufs directly to the pool in bulk + * if fast free offload is engaged + */ + if (!MLX5_TXOFF_CONFIG(MULTI) && txq->fast_free) { + mbuf = *pkts; + pool = mbuf->pool; + rte_mempool_put_bulk(pool, (void *)pkts, pkts_n); + return; + } + for (;;) { + for (;;) { + /* + * Decrement mbuf reference counter, detach + * indirect and external buffers if needed. + */ + mbuf = rte_pktmbuf_prefree_seg(*pkts); + if (likely(mbuf != NULL)) { + MLX5_ASSERT(mbuf == *pkts); + if (likely(n_free != 0)) { + if (unlikely(pool != mbuf->pool)) + /* From different pool. */ + break; + } else { + /* Start new scan array. */ + pool = mbuf->pool; + p_free = pkts; + } + ++n_free; + ++pkts; + --pkts_n; + if (unlikely(pkts_n == 0)) { + mbuf = NULL; + break; + } + } else { + /* + * This happens if mbuf is still referenced. + * We can't put it back to the pool, skip. + */ + ++pkts; + --pkts_n; + if (unlikely(n_free != 0)) + /* There is some array to free.*/ + break; + if (unlikely(pkts_n == 0)) + /* Last mbuf, nothing to free. */ + return; + } + } + for (;;) { + /* + * This loop is implemented to avoid multiple + * inlining of rte_mempool_put_bulk(). + */ + MLX5_ASSERT(pool); + MLX5_ASSERT(p_free); + MLX5_ASSERT(n_free); + /* + * Free the array of pre-freed mbufs + * belonging to the same memory pool. + */ + rte_mempool_put_bulk(pool, (void *)p_free, n_free); + if (unlikely(mbuf != NULL)) { + /* There is the request to start new scan. */ + pool = mbuf->pool; + p_free = pkts++; + n_free = 1; + --pkts_n; + if (likely(pkts_n != 0)) + break; + /* + * This is the last mbuf to be freed. + * Do one more loop iteration to complete. + * This is rare case of the last unique mbuf. + */ + mbuf = NULL; + continue; + } + if (likely(pkts_n == 0)) + return; + n_free = 0; + break; + } + } +} + +/** + * No inline version to free buffers for optimal call + * on the tx_burst completion. + */ +static __rte_noinline void +__mlx5_tx_free_mbuf(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + unsigned int olx __rte_unused) +{ + mlx5_tx_free_mbuf(txq, pkts, pkts_n, olx); +} + +/** + * Free the mbuf from the elts ring buffer till new tail. + * + * @param txq + * Pointer to Tx queue structure. + * @param tail + * Index in elts to free up to, becomes new elts tail. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_free_elts(struct mlx5_txq_data *__rte_restrict txq, + uint16_t tail, + unsigned int olx __rte_unused) +{ + uint16_t n_elts = tail - txq->elts_tail; + + MLX5_ASSERT(n_elts); + MLX5_ASSERT(n_elts <= txq->elts_s); + /* + * Implement a loop to support ring buffer wraparound + * with single inlining of mlx5_tx_free_mbuf(). + */ + do { + unsigned int part; + + part = txq->elts_s - (txq->elts_tail & txq->elts_m); + part = RTE_MIN(part, n_elts); + MLX5_ASSERT(part); + MLX5_ASSERT(part <= txq->elts_s); + mlx5_tx_free_mbuf(txq, + &txq->elts[txq->elts_tail & txq->elts_m], + part, olx); + txq->elts_tail += part; + n_elts -= part; + } while (n_elts); +} + +/** + * Store the mbuf being sent into elts ring buffer. + * On Tx completion these mbufs will be freed. + * + * @param txq + * Pointer to Tx queue structure. + * @param pkts + * Pointer to array of packets to be stored. + * @param pkts_n + * Number of packets to be stored. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_copy_elts(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + unsigned int olx __rte_unused) +{ + unsigned int part; + struct rte_mbuf **elts = (struct rte_mbuf **)txq->elts; + + MLX5_ASSERT(pkts); + MLX5_ASSERT(pkts_n); + part = txq->elts_s - (txq->elts_head & txq->elts_m); + MLX5_ASSERT(part); + MLX5_ASSERT(part <= txq->elts_s); + /* This code is a good candidate for vectorizing with SIMD. */ + rte_memcpy((void *)(elts + (txq->elts_head & txq->elts_m)), + (void *)pkts, + RTE_MIN(part, pkts_n) * sizeof(struct rte_mbuf *)); + txq->elts_head += pkts_n; + if (unlikely(part < pkts_n)) + /* The copy is wrapping around the elts array. */ + rte_memcpy((void *)elts, (void *)(pkts + part), + (pkts_n - part) * sizeof(struct rte_mbuf *)); +} + +/** + * Check if the completion request flag should be set in the last WQE. + * Both pushed mbufs and WQEs are monitored and the completion request + * flag is set if any of thresholds is reached. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_request_completion(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + uint16_t head = txq->elts_head; + unsigned int part; + + part = MLX5_TXOFF_CONFIG(INLINE) ? + 0 : loc->pkts_sent - loc->pkts_copy; + head += part; + if ((uint16_t)(head - txq->elts_comp) >= MLX5_TX_COMP_THRESH || + (MLX5_TXOFF_CONFIG(INLINE) && + (uint16_t)(txq->wqe_ci - txq->wqe_comp) >= txq->wqe_thres)) { + volatile struct mlx5_wqe *last = loc->wqe_last; + + MLX5_ASSERT(last); + txq->elts_comp = head; + if (MLX5_TXOFF_CONFIG(INLINE)) + txq->wqe_comp = txq->wqe_ci; + /* Request unconditional completion on last WQE. */ + last->cseg.flags = RTE_BE32(MLX5_COMP_ALWAYS << + MLX5_COMP_MODE_OFFSET); + /* Save elts_head in dedicated free on completion queue. */ +#ifdef RTE_LIBRTE_MLX5_DEBUG + txq->fcqs[txq->cq_pi++ & txq->cqe_m] = head | + (last->cseg.opcode >> 8) << 16; +#else + txq->fcqs[txq->cq_pi++ & txq->cqe_m] = head; +#endif + /* A CQE slot must always be available. */ + MLX5_ASSERT((txq->cq_pi - txq->cq_ci) <= txq->cqe_s); + } +} + +/** + * Build the Control Segment with specified opcode: + * - MLX5_OPCODE_SEND + * - MLX5_OPCODE_ENHANCED_MPSW + * - MLX5_OPCODE_TSO + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Control Segment. + * @param ds + * Supposed length of WQE in segments. + * @param opcode + * SQ WQE opcode to put into Control Segment. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_cseg_init(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc __rte_unused, + struct mlx5_wqe *__rte_restrict wqe, + unsigned int ds, + unsigned int opcode, + unsigned int olx __rte_unused) +{ + struct mlx5_wqe_cseg *__rte_restrict cs = &wqe->cseg; + + /* For legacy MPW replace the EMPW by TSO with modifier. */ + if (MLX5_TXOFF_CONFIG(MPW) && opcode == MLX5_OPCODE_ENHANCED_MPSW) + opcode = MLX5_OPCODE_TSO | MLX5_OPC_MOD_MPW << 24; + cs->opcode = rte_cpu_to_be_32((txq->wqe_ci << 8) | opcode); + cs->sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); + cs->flags = RTE_BE32(MLX5_COMP_ONLY_FIRST_ERR << + MLX5_COMP_MODE_OFFSET); + cs->misc = RTE_BE32(0); +} + +/** + * Build the Synchronize Queue Segment with specified completion index. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Control Segment. + * @param wci + * Completion index in Clock Queue to wait. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_wseg_init(struct mlx5_txq_data *restrict txq, + struct mlx5_txq_local *restrict loc __rte_unused, + struct mlx5_wqe *restrict wqe, + unsigned int wci, + unsigned int olx __rte_unused) +{ + struct mlx5_wqe_qseg *qs; + + qs = RTE_PTR_ADD(wqe, MLX5_WSEG_SIZE); + qs->max_index = rte_cpu_to_be_32(wci); + qs->qpn_cqn = rte_cpu_to_be_32(txq->sh->txpp.clock_queue.cq_obj.cq->id); + qs->reserved0 = RTE_BE32(0); + qs->reserved1 = RTE_BE32(0); +} + +/** + * Build the Ethernet Segment without inlined data. + * Supports Software Parser, Checksums and VLAN insertion Tx offload features. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Ethernet Segment. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_eseg_none(struct mlx5_txq_data *__rte_restrict txq __rte_unused, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe *__rte_restrict wqe, + unsigned int olx) +{ + struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; + uint32_t csum; + + /* + * Calculate and set check sum flags first, dword field + * in segment may be shared with Software Parser flags. + */ + csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; + es->flags = rte_cpu_to_le_32(csum); + /* + * Calculate and set Software Parser offsets and flags. + * These flags a set for custom UDP and IP tunnel packets. + */ + es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); + /* Fill metadata field if needed. */ + es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? + loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? + *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; + /* Engage VLAN tag insertion feature if requested. */ + if (MLX5_TXOFF_CONFIG(VLAN) && + loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { + /* + * We should get here only if device support + * this feature correctly. + */ + MLX5_ASSERT(txq->vlan_en); + es->inline_hdr = rte_cpu_to_be_32(MLX5_ETH_WQE_VLAN_INSERT | + loc->mbuf->vlan_tci); + } else { + es->inline_hdr = RTE_BE32(0); + } +} + +/** + * Build the Ethernet Segment with minimal inlined data + * of MLX5_ESEG_MIN_INLINE_SIZE bytes length. This is + * used to fill the gap in single WQEBB WQEs. + * Supports Software Parser, Checksums and VLAN + * insertion Tx offload features. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Ethernet Segment. + * @param vlan + * Length of VLAN tag insertion if any. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_eseg_dmin(struct mlx5_txq_data *__rte_restrict txq __rte_unused, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe *__rte_restrict wqe, + unsigned int vlan, + unsigned int olx) +{ + struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; + uint32_t csum; + uint8_t *psrc, *pdst; + + /* + * Calculate and set check sum flags first, dword field + * in segment may be shared with Software Parser flags. + */ + csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; + es->flags = rte_cpu_to_le_32(csum); + /* + * Calculate and set Software Parser offsets and flags. + * These flags a set for custom UDP and IP tunnel packets. + */ + es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); + /* Fill metadata field if needed. */ + es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? + loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? + *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; + psrc = rte_pktmbuf_mtod(loc->mbuf, uint8_t *); + es->inline_hdr_sz = RTE_BE16(MLX5_ESEG_MIN_INLINE_SIZE); + es->inline_data = *(unaligned_uint16_t *)psrc; + psrc += sizeof(uint16_t); + pdst = (uint8_t *)(es + 1); + if (MLX5_TXOFF_CONFIG(VLAN) && vlan) { + /* Implement VLAN tag insertion as part inline data. */ + memcpy(pdst, psrc, 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t)); + pdst += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); + psrc += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); + /* Insert VLAN ethertype + VLAN tag. */ + *(unaligned_uint32_t *)pdst = rte_cpu_to_be_32 + ((RTE_ETHER_TYPE_VLAN << 16) | + loc->mbuf->vlan_tci); + pdst += sizeof(struct rte_vlan_hdr); + /* Copy the rest two bytes from packet data. */ + MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, sizeof(uint16_t))); + *(uint16_t *)pdst = *(unaligned_uint16_t *)psrc; + } else { + /* Fill the gap in the title WQEBB with inline data. */ + rte_mov16(pdst, psrc); + } +} + +/** + * Build the Ethernet Segment with entire packet data inlining. Checks the + * boundary of WQEBB and ring buffer wrapping, supports Software Parser, + * Checksums and VLAN insertion Tx offload features. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Ethernet Segment. + * @param vlan + * Length of VLAN tag insertion if any. + * @param inlen + * Length of data to inline (VLAN included, if any). + * @param tso + * TSO flag, set mss field from the packet. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Pointer to the next Data Segment (aligned and wrapped around). + */ +static __rte_always_inline struct mlx5_wqe_dseg * +mlx5_tx_eseg_data(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe *__rte_restrict wqe, + unsigned int vlan, + unsigned int inlen, + unsigned int tso, + unsigned int olx) +{ + struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; + uint32_t csum; + uint8_t *psrc, *pdst; + unsigned int part; + + /* + * Calculate and set check sum flags first, dword field + * in segment may be shared with Software Parser flags. + */ + csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; + if (tso) { + csum <<= 24; + csum |= loc->mbuf->tso_segsz; + es->flags = rte_cpu_to_be_32(csum); + } else { + es->flags = rte_cpu_to_le_32(csum); + } + /* + * Calculate and set Software Parser offsets and flags. + * These flags a set for custom UDP and IP tunnel packets. + */ + es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); + /* Fill metadata field if needed. */ + es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? + loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? + *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; + psrc = rte_pktmbuf_mtod(loc->mbuf, uint8_t *); + es->inline_hdr_sz = rte_cpu_to_be_16(inlen); + es->inline_data = *(unaligned_uint16_t *)psrc; + psrc += sizeof(uint16_t); + pdst = (uint8_t *)(es + 1); + if (MLX5_TXOFF_CONFIG(VLAN) && vlan) { + /* Implement VLAN tag insertion as part inline data. */ + memcpy(pdst, psrc, 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t)); + pdst += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); + psrc += 2 * RTE_ETHER_ADDR_LEN - sizeof(uint16_t); + /* Insert VLAN ethertype + VLAN tag. */ + *(unaligned_uint32_t *)pdst = rte_cpu_to_be_32 + ((RTE_ETHER_TYPE_VLAN << 16) | + loc->mbuf->vlan_tci); + pdst += sizeof(struct rte_vlan_hdr); + /* Copy the rest two bytes from packet data. */ + MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, sizeof(uint16_t))); + *(uint16_t *)pdst = *(unaligned_uint16_t *)psrc; + psrc += sizeof(uint16_t); + } else { + /* Fill the gap in the title WQEBB with inline data. */ + rte_mov16(pdst, psrc); + psrc += sizeof(rte_v128u32_t); + } + pdst = (uint8_t *)(es + 2); + MLX5_ASSERT(inlen >= MLX5_ESEG_MIN_INLINE_SIZE); + MLX5_ASSERT(pdst < (uint8_t *)txq->wqes_end); + inlen -= MLX5_ESEG_MIN_INLINE_SIZE; + if (!inlen) { + MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE)); + return (struct mlx5_wqe_dseg *)pdst; + } + /* + * The WQEBB space availability is checked by caller. + * Here we should be aware of WQE ring buffer wraparound only. + */ + part = (uint8_t *)txq->wqes_end - pdst; + part = RTE_MIN(part, inlen); + do { + rte_memcpy(pdst, psrc, part); + inlen -= part; + if (likely(!inlen)) { + /* + * If return value is not used by the caller + * the code below will be optimized out. + */ + pdst += part; + pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); + if (unlikely(pdst >= (uint8_t *)txq->wqes_end)) + pdst = (uint8_t *)txq->wqes; + return (struct mlx5_wqe_dseg *)pdst; + } + pdst = (uint8_t *)txq->wqes; + psrc += part; + part = inlen; + } while (true); +} + +/** + * Copy data from chain of mbuf to the specified linear buffer. + * Checksums and VLAN insertion Tx offload features. If data + * from some mbuf copied completely this mbuf is freed. Local + * structure is used to keep the byte stream state. + * + * @param pdst + * Pointer to the destination linear buffer. + * @param loc + * Pointer to burst routine local context. + * @param len + * Length of data to be copied. + * @param must + * Length of data to be copied ignoring no inline hint. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Number of actual copied data bytes. This is always greater than or + * equal to must parameter and might be lesser than len in no inline + * hint flag is encountered. + */ +static __rte_always_inline unsigned int +mlx5_tx_mseg_memcpy(uint8_t *pdst, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int len, + unsigned int must, + unsigned int olx __rte_unused) +{ + struct rte_mbuf *mbuf; + unsigned int part, dlen, copy = 0; + uint8_t *psrc; + + MLX5_ASSERT(len); + MLX5_ASSERT(must <= len); + do { + /* Allow zero length packets, must check first. */ + dlen = rte_pktmbuf_data_len(loc->mbuf); + if (dlen <= loc->mbuf_off) { + /* Exhausted packet, just free. */ + mbuf = loc->mbuf; + loc->mbuf = mbuf->next; + rte_pktmbuf_free_seg(mbuf); + loc->mbuf_off = 0; + MLX5_ASSERT(loc->mbuf_nseg > 1); + MLX5_ASSERT(loc->mbuf); + --loc->mbuf_nseg; + if (loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE) { + unsigned int diff; + + if (copy >= must) { + /* + * We already copied the minimal + * requested amount of data. + */ + return copy; + } + diff = must - copy; + if (diff <= rte_pktmbuf_data_len(loc->mbuf)) { + /* + * Copy only the minimal required + * part of the data buffer. + */ + len = diff; + } + } + continue; + } + dlen -= loc->mbuf_off; + psrc = rte_pktmbuf_mtod_offset(loc->mbuf, uint8_t *, + loc->mbuf_off); + part = RTE_MIN(len, dlen); + rte_memcpy(pdst, psrc, part); + copy += part; + loc->mbuf_off += part; + len -= part; + if (!len) { + if (loc->mbuf_off >= rte_pktmbuf_data_len(loc->mbuf)) { + loc->mbuf_off = 0; + /* Exhausted packet, just free. */ + mbuf = loc->mbuf; + loc->mbuf = mbuf->next; + rte_pktmbuf_free_seg(mbuf); + loc->mbuf_off = 0; + MLX5_ASSERT(loc->mbuf_nseg >= 1); + --loc->mbuf_nseg; + } + return copy; + } + pdst += part; + } while (true); +} + +/** + * Build the Ethernet Segment with inlined data from multi-segment packet. + * Checks the boundary of WQEBB and ring buffer wrapping, supports Software + * Parser, Checksums and VLAN insertion Tx offload features. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Ethernet Segment. + * @param vlan + * Length of VLAN tag insertion if any. + * @param inlen + * Length of data to inline (VLAN included, if any). + * @param tso + * TSO flag, set mss field from the packet. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Pointer to the next Data Segment (aligned and possible NOT wrapped + * around - caller should do wrapping check on its own). + */ +static __rte_always_inline struct mlx5_wqe_dseg * +mlx5_tx_eseg_mdat(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe *__rte_restrict wqe, + unsigned int vlan, + unsigned int inlen, + unsigned int tso, + unsigned int olx) +{ + struct mlx5_wqe_eseg *__rte_restrict es = &wqe->eseg; + uint32_t csum; + uint8_t *pdst; + unsigned int part, tlen = 0; + + /* + * Calculate and set check sum flags first, uint32_t field + * in segment may be shared with Software Parser flags. + */ + csum = MLX5_TXOFF_CONFIG(CSUM) ? txq_ol_cksum_to_cs(loc->mbuf) : 0; + if (tso) { + csum <<= 24; + csum |= loc->mbuf->tso_segsz; + es->flags = rte_cpu_to_be_32(csum); + } else { + es->flags = rte_cpu_to_le_32(csum); + } + /* + * Calculate and set Software Parser offsets and flags. + * These flags a set for custom UDP and IP tunnel packets. + */ + es->swp_offs = txq_mbuf_to_swp(loc, &es->swp_flags, olx); + /* Fill metadata field if needed. */ + es->metadata = MLX5_TXOFF_CONFIG(METADATA) ? + loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? + *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0 : 0; + MLX5_ASSERT(inlen >= MLX5_ESEG_MIN_INLINE_SIZE); + pdst = (uint8_t *)&es->inline_data; + if (MLX5_TXOFF_CONFIG(VLAN) && vlan) { + /* Implement VLAN tag insertion as part inline data. */ + mlx5_tx_mseg_memcpy(pdst, loc, + 2 * RTE_ETHER_ADDR_LEN, + 2 * RTE_ETHER_ADDR_LEN, olx); + pdst += 2 * RTE_ETHER_ADDR_LEN; + *(unaligned_uint32_t *)pdst = rte_cpu_to_be_32 + ((RTE_ETHER_TYPE_VLAN << 16) | + loc->mbuf->vlan_tci); + pdst += sizeof(struct rte_vlan_hdr); + tlen += 2 * RTE_ETHER_ADDR_LEN + sizeof(struct rte_vlan_hdr); + } + MLX5_ASSERT(pdst < (uint8_t *)txq->wqes_end); + /* + * The WQEBB space availability is checked by caller. + * Here we should be aware of WQE ring buffer wraparound only. + */ + part = (uint8_t *)txq->wqes_end - pdst; + part = RTE_MIN(part, inlen - tlen); + MLX5_ASSERT(part); + do { + unsigned int copy; + + /* + * Copying may be interrupted inside the routine + * if run into no inline hint flag. + */ + copy = tlen >= txq->inlen_mode ? 0 : (txq->inlen_mode - tlen); + copy = mlx5_tx_mseg_memcpy(pdst, loc, part, copy, olx); + tlen += copy; + if (likely(inlen <= tlen) || copy < part) { + es->inline_hdr_sz = rte_cpu_to_be_16(tlen); + pdst += copy; + pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); + return (struct mlx5_wqe_dseg *)pdst; + } + pdst = (uint8_t *)txq->wqes; + part = inlen - tlen; + } while (true); +} + +/** + * Build the Data Segment of pointer type. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param dseg + * Pointer to WQE to fill with built Data Segment. + * @param buf + * Data buffer to point. + * @param len + * Data buffer length. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_dseg_ptr(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe_dseg *__rte_restrict dseg, + uint8_t *buf, + unsigned int len, + unsigned int olx __rte_unused) + +{ + MLX5_ASSERT(len); + dseg->bcount = rte_cpu_to_be_32(len); + dseg->lkey = mlx5_tx_mb2mr(txq, loc->mbuf); + dseg->pbuf = rte_cpu_to_be_64((uintptr_t)buf); +} + +/** + * Build the Data Segment of pointer type or inline if data length is less than + * buffer in minimal Data Segment size. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param dseg + * Pointer to WQE to fill with built Data Segment. + * @param buf + * Data buffer to point. + * @param len + * Data buffer length. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_dseg_iptr(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe_dseg *__rte_restrict dseg, + uint8_t *buf, + unsigned int len, + unsigned int olx __rte_unused) + +{ + uintptr_t dst, src; + + MLX5_ASSERT(len); + if (len > MLX5_DSEG_MIN_INLINE_SIZE) { + dseg->bcount = rte_cpu_to_be_32(len); + dseg->lkey = mlx5_tx_mb2mr(txq, loc->mbuf); + dseg->pbuf = rte_cpu_to_be_64((uintptr_t)buf); + + return; + } + dseg->bcount = rte_cpu_to_be_32(len | MLX5_ETH_WQE_DATA_INLINE); + /* Unrolled implementation of generic rte_memcpy. */ + dst = (uintptr_t)&dseg->inline_data[0]; + src = (uintptr_t)buf; + if (len & 0x08) { +#ifdef RTE_ARCH_STRICT_ALIGN + MLX5_ASSERT(dst == RTE_PTR_ALIGN(dst, sizeof(uint32_t))); + *(uint32_t *)dst = *(unaligned_uint32_t *)src; + dst += sizeof(uint32_t); + src += sizeof(uint32_t); + *(uint32_t *)dst = *(unaligned_uint32_t *)src; + dst += sizeof(uint32_t); + src += sizeof(uint32_t); +#else + *(uint64_t *)dst = *(unaligned_uint64_t *)src; + dst += sizeof(uint64_t); + src += sizeof(uint64_t); +#endif + } + if (len & 0x04) { + *(uint32_t *)dst = *(unaligned_uint32_t *)src; + dst += sizeof(uint32_t); + src += sizeof(uint32_t); + } + if (len & 0x02) { + *(uint16_t *)dst = *(unaligned_uint16_t *)src; + dst += sizeof(uint16_t); + src += sizeof(uint16_t); + } + if (len & 0x01) + *(uint8_t *)dst = *(uint8_t *)src; +} + +/** + * Build the Data Segment of inlined data from single + * segment packet, no VLAN insertion. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param dseg + * Pointer to WQE to fill with built Data Segment. + * @param buf + * Data buffer to point. + * @param len + * Data buffer length. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Pointer to the next Data Segment after inlined data. + * Ring buffer wraparound check is needed. We do not do it here because it + * may not be needed for the last packet in the eMPW session. + */ +static __rte_always_inline struct mlx5_wqe_dseg * +mlx5_tx_dseg_empw(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc __rte_unused, + struct mlx5_wqe_dseg *__rte_restrict dseg, + uint8_t *buf, + unsigned int len, + unsigned int olx __rte_unused) +{ + unsigned int part; + uint8_t *pdst; + + if (!MLX5_TXOFF_CONFIG(MPW)) { + /* Store the descriptor byte counter for eMPW sessions. */ + dseg->bcount = rte_cpu_to_be_32(len | MLX5_ETH_WQE_DATA_INLINE); + pdst = &dseg->inline_data[0]; + } else { + /* The entire legacy MPW session counter is stored on close. */ + pdst = (uint8_t *)dseg; + } + /* + * The WQEBB space availability is checked by caller. + * Here we should be aware of WQE ring buffer wraparound only. + */ + part = (uint8_t *)txq->wqes_end - pdst; + part = RTE_MIN(part, len); + do { + rte_memcpy(pdst, buf, part); + len -= part; + if (likely(!len)) { + pdst += part; + if (!MLX5_TXOFF_CONFIG(MPW)) + pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); + /* Note: no final wraparound check here. */ + return (struct mlx5_wqe_dseg *)pdst; + } + pdst = (uint8_t *)txq->wqes; + buf += part; + part = len; + } while (true); +} + +/** + * Build the Data Segment of inlined data from single + * segment packet with VLAN insertion. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param dseg + * Pointer to the dseg fill with built Data Segment. + * @param buf + * Data buffer to point. + * @param len + * Data buffer length. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Pointer to the next Data Segment after inlined data. + * Ring buffer wraparound check is needed. + */ +static __rte_always_inline struct mlx5_wqe_dseg * +mlx5_tx_dseg_vlan(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc __rte_unused, + struct mlx5_wqe_dseg *__rte_restrict dseg, + uint8_t *buf, + unsigned int len, + unsigned int olx __rte_unused) + +{ + unsigned int part; + uint8_t *pdst; + + MLX5_ASSERT(len > MLX5_ESEG_MIN_INLINE_SIZE); + if (!MLX5_TXOFF_CONFIG(MPW)) { + /* Store the descriptor byte counter for eMPW sessions. */ + dseg->bcount = rte_cpu_to_be_32 + ((len + sizeof(struct rte_vlan_hdr)) | + MLX5_ETH_WQE_DATA_INLINE); + pdst = &dseg->inline_data[0]; + } else { + /* The entire legacy MPW session counter is stored on close. */ + pdst = (uint8_t *)dseg; + } + memcpy(pdst, buf, MLX5_DSEG_MIN_INLINE_SIZE); + buf += MLX5_DSEG_MIN_INLINE_SIZE; + pdst += MLX5_DSEG_MIN_INLINE_SIZE; + len -= MLX5_DSEG_MIN_INLINE_SIZE; + /* Insert VLAN ethertype + VLAN tag. Pointer is aligned. */ + MLX5_ASSERT(pdst == RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE)); + if (unlikely(pdst >= (uint8_t *)txq->wqes_end)) + pdst = (uint8_t *)txq->wqes; + *(uint32_t *)pdst = rte_cpu_to_be_32((RTE_ETHER_TYPE_VLAN << 16) | + loc->mbuf->vlan_tci); + pdst += sizeof(struct rte_vlan_hdr); + /* + * The WQEBB space availability is checked by caller. + * Here we should be aware of WQE ring buffer wraparound only. + */ + part = (uint8_t *)txq->wqes_end - pdst; + part = RTE_MIN(part, len); + do { + rte_memcpy(pdst, buf, part); + len -= part; + if (likely(!len)) { + pdst += part; + if (!MLX5_TXOFF_CONFIG(MPW)) + pdst = RTE_PTR_ALIGN(pdst, MLX5_WSEG_SIZE); + /* Note: no final wraparound check here. */ + return (struct mlx5_wqe_dseg *)pdst; + } + pdst = (uint8_t *)txq->wqes; + buf += part; + part = len; + } while (true); +} + +/** + * Build the Ethernet Segment with optionally inlined data with + * VLAN insertion and following Data Segments (if any) from + * multi-segment packet. Used by ordinary send and TSO. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param wqe + * Pointer to WQE to fill with built Ethernet/Data Segments. + * @param vlan + * Length of VLAN header to insert, 0 means no VLAN insertion. + * @param inlen + * Data length to inline. For TSO this parameter specifies exact value, + * for ordinary send routine can be aligned by caller to provide better WQE + * space saving and data buffer start address alignment. + * This length includes VLAN header being inserted. + * @param tso + * Zero means ordinary send, inlined data can be extended, + * otherwise this is TSO, inlined data length is fixed. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * Actual size of built WQE in segments. + */ +static __rte_always_inline unsigned int +mlx5_tx_mseg_build(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + struct mlx5_wqe *__rte_restrict wqe, + unsigned int vlan, + unsigned int inlen, + unsigned int tso, + unsigned int olx __rte_unused) +{ + struct mlx5_wqe_dseg *__rte_restrict dseg; + unsigned int ds; + + MLX5_ASSERT((rte_pktmbuf_pkt_len(loc->mbuf) + vlan) >= inlen); + loc->mbuf_nseg = NB_SEGS(loc->mbuf); + loc->mbuf_off = 0; + + dseg = mlx5_tx_eseg_mdat(txq, loc, wqe, vlan, inlen, tso, olx); + if (!loc->mbuf_nseg) + goto dseg_done; + /* + * There are still some mbuf remaining, not inlined. + * The first mbuf may be partially inlined and we + * must process the possible non-zero data offset. + */ + if (loc->mbuf_off) { + unsigned int dlen; + uint8_t *dptr; + + /* + * Exhausted packets must be dropped before. + * Non-zero offset means there are some data + * remained in the packet. + */ + MLX5_ASSERT(loc->mbuf_off < rte_pktmbuf_data_len(loc->mbuf)); + MLX5_ASSERT(rte_pktmbuf_data_len(loc->mbuf)); + dptr = rte_pktmbuf_mtod_offset(loc->mbuf, uint8_t *, + loc->mbuf_off); + dlen = rte_pktmbuf_data_len(loc->mbuf) - loc->mbuf_off; + /* + * Build the pointer/minimal Data Segment. + * Do ring buffer wrapping check in advance. + */ + if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) + dseg = (struct mlx5_wqe_dseg *)txq->wqes; + mlx5_tx_dseg_iptr(txq, loc, dseg, dptr, dlen, olx); + /* Store the mbuf to be freed on completion. */ + MLX5_ASSERT(loc->elts_free); + txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; + --loc->elts_free; + ++dseg; + if (--loc->mbuf_nseg == 0) + goto dseg_done; + loc->mbuf = loc->mbuf->next; + loc->mbuf_off = 0; + } + do { + if (unlikely(!rte_pktmbuf_data_len(loc->mbuf))) { + struct rte_mbuf *mbuf; + + /* Zero length segment found, just skip. */ + mbuf = loc->mbuf; + loc->mbuf = loc->mbuf->next; + rte_pktmbuf_free_seg(mbuf); + if (--loc->mbuf_nseg == 0) + break; + } else { + if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) + dseg = (struct mlx5_wqe_dseg *)txq->wqes; + mlx5_tx_dseg_iptr + (txq, loc, dseg, + rte_pktmbuf_mtod(loc->mbuf, uint8_t *), + rte_pktmbuf_data_len(loc->mbuf), olx); + MLX5_ASSERT(loc->elts_free); + txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; + --loc->elts_free; + ++dseg; + if (--loc->mbuf_nseg == 0) + break; + loc->mbuf = loc->mbuf->next; + } + } while (true); + +dseg_done: + /* Calculate actual segments used from the dseg pointer. */ + if ((uintptr_t)wqe < (uintptr_t)dseg) + ds = ((uintptr_t)dseg - (uintptr_t)wqe) / MLX5_WSEG_SIZE; + else + ds = (((uintptr_t)dseg - (uintptr_t)wqe) + + txq->wqe_s * MLX5_WQE_SIZE) / MLX5_WSEG_SIZE; + return ds; +} + +/** + * The routine checks timestamp flag in the current packet, + * and push WAIT WQE into the queue if scheduling is required. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_SINGLE - continue processing with the packet. + * MLX5_TXCMP_CODE_MULTI - the WAIT inserted, continue processing. + * Local context variables partially updated. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_schedule_send(struct mlx5_txq_data *restrict txq, + struct mlx5_txq_local *restrict loc, + unsigned int olx) +{ + if (MLX5_TXOFF_CONFIG(TXPP) && + loc->mbuf->ol_flags & txq->ts_mask) { + struct mlx5_wqe *wqe; + uint64_t ts; + int32_t wci; + + /* + * Estimate the required space quickly and roughly. + * We would like to ensure the packet can be pushed + * to the queue and we won't get the orphan WAIT WQE. + */ + if (loc->wqe_free <= MLX5_WQE_SIZE_MAX / MLX5_WQE_SIZE || + loc->elts_free < NB_SEGS(loc->mbuf)) + return MLX5_TXCMP_CODE_EXIT; + /* Convert the timestamp into completion to wait. */ + ts = *RTE_MBUF_DYNFIELD(loc->mbuf, txq->ts_offset, uint64_t *); + wci = mlx5_txpp_convert_tx_ts(txq->sh, ts); + if (unlikely(wci < 0)) + return MLX5_TXCMP_CODE_SINGLE; + /* Build the WAIT WQE with specified completion. */ + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + mlx5_tx_cseg_init(txq, loc, wqe, 2, MLX5_OPCODE_WAIT, olx); + mlx5_tx_wseg_init(txq, loc, wqe, wci, olx); + ++txq->wqe_ci; + --loc->wqe_free; + return MLX5_TXCMP_CODE_MULTI; + } + return MLX5_TXCMP_CODE_SINGLE; +} + +/** + * Tx one packet function for multi-segment TSO. Supports all + * types of Tx offloads, uses MLX5_OPCODE_TSO to build WQEs, + * sends one packet per WQE. + * + * This routine is responsible for storing processed mbuf + * into elts ring buffer and update elts_head. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. + * Local context variables partially updated. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_packet_multi_tso(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + struct mlx5_wqe *__rte_restrict wqe; + unsigned int ds, dlen, inlen, ntcp, vlan = 0; + + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + /* + * Calculate data length to be inlined to estimate + * the required space in WQE ring buffer. + */ + dlen = rte_pktmbuf_pkt_len(loc->mbuf); + if (MLX5_TXOFF_CONFIG(VLAN) && loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) + vlan = sizeof(struct rte_vlan_hdr); + inlen = loc->mbuf->l2_len + vlan + + loc->mbuf->l3_len + loc->mbuf->l4_len; + if (unlikely((!inlen || !loc->mbuf->tso_segsz))) + return MLX5_TXCMP_CODE_ERROR; + if (loc->mbuf->ol_flags & PKT_TX_TUNNEL_MASK) + inlen += loc->mbuf->outer_l2_len + loc->mbuf->outer_l3_len; + /* Packet must contain all TSO headers. */ + if (unlikely(inlen > MLX5_MAX_TSO_HEADER || + inlen <= MLX5_ESEG_MIN_INLINE_SIZE || + inlen > (dlen + vlan))) + return MLX5_TXCMP_CODE_ERROR; + MLX5_ASSERT(inlen >= txq->inlen_mode); + /* + * Check whether there are enough free WQEBBs: + * - Control Segment + * - Ethernet Segment + * - First Segment of inlined Ethernet data + * - ... data continued ... + * - Data Segments of pointer/min inline type + */ + ds = NB_SEGS(loc->mbuf) + 2 + (inlen - + MLX5_ESEG_MIN_INLINE_SIZE + + MLX5_WSEG_SIZE + + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; + if (unlikely(loc->wqe_free < ((ds + 3) / 4))) + return MLX5_TXCMP_CODE_EXIT; + /* Check for maximal WQE size. */ + if (unlikely((MLX5_WQE_SIZE_MAX / MLX5_WSEG_SIZE) < ((ds + 3) / 4))) + return MLX5_TXCMP_CODE_ERROR; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes/packets counters. */ + ntcp = (dlen - (inlen - vlan) + loc->mbuf->tso_segsz - 1) / + loc->mbuf->tso_segsz; + /* + * One will be added for mbuf itself at the end of the mlx5_tx_burst + * from loc->pkts_sent field. + */ + --ntcp; + txq->stats.opackets += ntcp; + txq->stats.obytes += dlen + vlan + ntcp * inlen; +#endif + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, 0, MLX5_OPCODE_TSO, olx); + ds = mlx5_tx_mseg_build(txq, loc, wqe, vlan, inlen, 1, olx); + wqe->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); + txq->wqe_ci += (ds + 3) / 4; + loc->wqe_free -= (ds + 3) / 4; + return MLX5_TXCMP_CODE_MULTI; +} + +/** + * Tx one packet function for multi-segment SEND. Supports all types of Tx + * offloads, uses MLX5_OPCODE_SEND to build WQEs, sends one packet per WQE, + * without any data inlining in Ethernet Segment. + * + * This routine is responsible for storing processed mbuf + * into elts ring buffer and update elts_head. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. + * Local context variables partially updated. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_packet_multi_send(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + struct mlx5_wqe_dseg *__rte_restrict dseg; + struct mlx5_wqe *__rte_restrict wqe; + unsigned int ds, nseg; + + MLX5_ASSERT(NB_SEGS(loc->mbuf) > 1); + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + /* + * No inline at all, it means the CPU cycles saving is prioritized at + * configuration, we should not copy any packet data to WQE. + */ + nseg = NB_SEGS(loc->mbuf); + ds = 2 + nseg; + if (unlikely(loc->wqe_free < ((ds + 3) / 4))) + return MLX5_TXCMP_CODE_EXIT; + /* Check for maximal WQE size. */ + if (unlikely((MLX5_WQE_SIZE_MAX / MLX5_WSEG_SIZE) < ((ds + 3) / 4))) + return MLX5_TXCMP_CODE_ERROR; + /* + * Some Tx offloads may cause an error if packet is not long enough, + * check against assumed minimal length. + */ + if (rte_pktmbuf_pkt_len(loc->mbuf) <= MLX5_ESEG_MIN_INLINE_SIZE) + return MLX5_TXCMP_CODE_ERROR; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + txq->stats.obytes += rte_pktmbuf_pkt_len(loc->mbuf); + if (MLX5_TXOFF_CONFIG(VLAN) && + loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) + txq->stats.obytes += sizeof(struct rte_vlan_hdr); +#endif + /* + * SEND WQE, one WQEBB: + * - Control Segment, SEND opcode + * - Ethernet Segment, optional VLAN, no inline + * - Data Segments, pointer only type + */ + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, ds, MLX5_OPCODE_SEND, olx); + mlx5_tx_eseg_none(txq, loc, wqe, olx); + dseg = &wqe->dseg[0]; + do { + if (unlikely(!rte_pktmbuf_data_len(loc->mbuf))) { + struct rte_mbuf *mbuf; + + /* + * Zero length segment found, have to correct total + * size of WQE in segments. + * It is supposed to be rare occasion, so in normal + * case (no zero length segments) we avoid extra + * writing to the Control Segment. + */ + --ds; + wqe->cseg.sq_ds -= RTE_BE32(1); + mbuf = loc->mbuf; + loc->mbuf = mbuf->next; + rte_pktmbuf_free_seg(mbuf); + if (--nseg == 0) + break; + } else { + mlx5_tx_dseg_ptr + (txq, loc, dseg, + rte_pktmbuf_mtod(loc->mbuf, uint8_t *), + rte_pktmbuf_data_len(loc->mbuf), olx); + txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; + --loc->elts_free; + if (--nseg == 0) + break; + ++dseg; + if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) + dseg = (struct mlx5_wqe_dseg *)txq->wqes; + loc->mbuf = loc->mbuf->next; + } + } while (true); + txq->wqe_ci += (ds + 3) / 4; + loc->wqe_free -= (ds + 3) / 4; + return MLX5_TXCMP_CODE_MULTI; +} + +/** + * Tx one packet function for multi-segment SEND. Supports all + * types of Tx offloads, uses MLX5_OPCODE_SEND to build WQEs, + * sends one packet per WQE, with data inlining in + * Ethernet Segment and minimal Data Segments. + * + * This routine is responsible for storing processed mbuf + * into elts ring buffer and update elts_head. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. + * Local context variables partially updated. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_packet_multi_inline(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + struct mlx5_wqe *__rte_restrict wqe; + unsigned int ds, inlen, dlen, vlan = 0; + + MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); + MLX5_ASSERT(NB_SEGS(loc->mbuf) > 1); + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + /* + * First calculate data length to be inlined + * to estimate the required space for WQE. + */ + dlen = rte_pktmbuf_pkt_len(loc->mbuf); + if (MLX5_TXOFF_CONFIG(VLAN) && loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) + vlan = sizeof(struct rte_vlan_hdr); + inlen = dlen + vlan; + /* Check against minimal length. */ + if (inlen <= MLX5_ESEG_MIN_INLINE_SIZE) + return MLX5_TXCMP_CODE_ERROR; + MLX5_ASSERT(txq->inlen_send >= MLX5_ESEG_MIN_INLINE_SIZE); + if (inlen > txq->inlen_send || + loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE) { + struct rte_mbuf *mbuf; + unsigned int nxlen; + uintptr_t start; + + /* + * Packet length exceeds the allowed inline data length, + * check whether the minimal inlining is required. + */ + if (txq->inlen_mode) { + MLX5_ASSERT(txq->inlen_mode >= + MLX5_ESEG_MIN_INLINE_SIZE); + MLX5_ASSERT(txq->inlen_mode <= txq->inlen_send); + inlen = txq->inlen_mode; + } else { + if (loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE || + !vlan || txq->vlan_en) { + /* + * VLAN insertion will be done inside by HW. + * It is not utmost effective - VLAN flag is + * checked twice, but we should proceed the + * inlining length correctly and take into + * account the VLAN header being inserted. + */ + return mlx5_tx_packet_multi_send + (txq, loc, olx); + } + inlen = MLX5_ESEG_MIN_INLINE_SIZE; + } + /* + * Now we know the minimal amount of data is requested + * to inline. Check whether we should inline the buffers + * from the chain beginning to eliminate some mbufs. + */ + mbuf = loc->mbuf; + nxlen = rte_pktmbuf_data_len(mbuf); + if (unlikely(nxlen <= txq->inlen_send)) { + /* We can inline first mbuf at least. */ + if (nxlen < inlen) { + unsigned int smlen; + + /* Scan mbufs till inlen filled. */ + do { + smlen = nxlen; + mbuf = NEXT(mbuf); + MLX5_ASSERT(mbuf); + nxlen = rte_pktmbuf_data_len(mbuf); + nxlen += smlen; + } while (unlikely(nxlen < inlen)); + if (unlikely(nxlen > txq->inlen_send)) { + /* We cannot inline entire mbuf. */ + smlen = inlen - smlen; + start = rte_pktmbuf_mtod_offset + (mbuf, uintptr_t, smlen); + goto do_align; + } + } + do { + inlen = nxlen; + mbuf = NEXT(mbuf); + /* There should be not end of packet. */ + MLX5_ASSERT(mbuf); + nxlen = inlen + rte_pktmbuf_data_len(mbuf); + } while (unlikely(nxlen < txq->inlen_send)); + } + start = rte_pktmbuf_mtod(mbuf, uintptr_t); + /* + * Check whether we can do inline to align start + * address of data buffer to cacheline. + */ +do_align: + start = (~start + 1) & (RTE_CACHE_LINE_SIZE - 1); + if (unlikely(start)) { + start += inlen; + if (start <= txq->inlen_send) + inlen = start; + } + } + /* + * Check whether there are enough free WQEBBs: + * - Control Segment + * - Ethernet Segment + * - First Segment of inlined Ethernet data + * - ... data continued ... + * - Data Segments of pointer/min inline type + * + * Estimate the number of Data Segments conservatively, + * supposing no any mbufs is being freed during inlining. + */ + MLX5_ASSERT(inlen <= txq->inlen_send); + ds = NB_SEGS(loc->mbuf) + 2 + (inlen - + MLX5_ESEG_MIN_INLINE_SIZE + + MLX5_WSEG_SIZE + + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; + if (unlikely(loc->wqe_free < ((ds + 3) / 4))) + return MLX5_TXCMP_CODE_EXIT; + /* Check for maximal WQE size. */ + if (unlikely((MLX5_WQE_SIZE_MAX / MLX5_WSEG_SIZE) < ((ds + 3) / 4))) + return MLX5_TXCMP_CODE_ERROR; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes/packets counters. */ + txq->stats.obytes += dlen + vlan; +#endif + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, 0, MLX5_OPCODE_SEND, olx); + ds = mlx5_tx_mseg_build(txq, loc, wqe, vlan, inlen, 0, olx); + wqe->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); + txq->wqe_ci += (ds + 3) / 4; + loc->wqe_free -= (ds + 3) / 4; + return MLX5_TXCMP_CODE_MULTI; +} + +/** + * Tx burst function for multi-segment packets. Supports all + * types of Tx offloads, uses MLX5_OPCODE_SEND/TSO to build WQEs, + * sends one packet per WQE. Function stops sending if it + * encounters the single-segment packet. + * + * This routine is responsible for storing processed mbuf + * into elts ring buffer and update elts_head. + * + * @param txq + * Pointer to TX queue structure. + * @param[in] pkts + * Packets to transmit. + * @param pkts_n + * Number of packets in array. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. + * MLX5_TXCMP_CODE_SINGLE - single-segment packet encountered. + * MLX5_TXCMP_CODE_TSO - TSO single-segment packet encountered. + * Local context variables updated. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_burst_mseg(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + MLX5_ASSERT(loc->elts_free && loc->wqe_free); + MLX5_ASSERT(pkts_n > loc->pkts_sent); + pkts += loc->pkts_sent + 1; + pkts_n -= loc->pkts_sent; + for (;;) { + enum mlx5_txcmp_code ret; + + MLX5_ASSERT(NB_SEGS(loc->mbuf) > 1); + /* + * Estimate the number of free elts quickly but conservatively. + * Some segment may be fully inlined and freed, + * ignore this here - precise estimation is costly. + */ + if (loc->elts_free < NB_SEGS(loc->mbuf)) + return MLX5_TXCMP_CODE_EXIT; + if (MLX5_TXOFF_CONFIG(TSO) && + unlikely(loc->mbuf->ol_flags & PKT_TX_TCP_SEG)) { + /* Proceed with multi-segment TSO. */ + ret = mlx5_tx_packet_multi_tso(txq, loc, olx); + } else if (MLX5_TXOFF_CONFIG(INLINE)) { + /* Proceed with multi-segment SEND with inlining. */ + ret = mlx5_tx_packet_multi_inline(txq, loc, olx); + } else { + /* Proceed with multi-segment SEND w/o inlining. */ + ret = mlx5_tx_packet_multi_send(txq, loc, olx); + } + if (ret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (ret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + /* WQE is built, go to the next packet. */ + ++loc->pkts_sent; + --pkts_n; + if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + loc->mbuf = *pkts++; + if (pkts_n > 1) + rte_prefetch0(*pkts); + if (likely(NB_SEGS(loc->mbuf) > 1)) + continue; + /* Here ends the series of multi-segment packets. */ + if (MLX5_TXOFF_CONFIG(TSO) && + unlikely(loc->mbuf->ol_flags & PKT_TX_TCP_SEG)) + return MLX5_TXCMP_CODE_TSO; + return MLX5_TXCMP_CODE_SINGLE; + } + MLX5_ASSERT(false); +} + +/** + * Tx burst function for single-segment packets with TSO. + * Supports all types of Tx offloads, except multi-packets. + * Uses MLX5_OPCODE_TSO to build WQEs, sends one packet per WQE. + * Function stops sending if it encounters the multi-segment + * packet or packet without TSO requested. + * + * The routine is responsible for storing processed mbuf into elts ring buffer + * and update elts_head if inline offloads is requested due to possible early + * freeing of the inlined mbufs (can not store pkts array in elts as a batch). + * + * @param txq + * Pointer to TX queue structure. + * @param[in] pkts + * Packets to transmit. + * @param pkts_n + * Number of packets in array. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. + * MLX5_TXCMP_CODE_SINGLE - single-segment packet encountered. + * MLX5_TXCMP_CODE_MULTI - multi-segment packet encountered. + * Local context variables updated. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_burst_tso(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + MLX5_ASSERT(loc->elts_free && loc->wqe_free); + MLX5_ASSERT(pkts_n > loc->pkts_sent); + pkts += loc->pkts_sent + 1; + pkts_n -= loc->pkts_sent; + for (;;) { + struct mlx5_wqe_dseg *__rte_restrict dseg; + struct mlx5_wqe *__rte_restrict wqe; + unsigned int ds, dlen, hlen, ntcp, vlan = 0; + uint8_t *dptr; + + MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + dlen = rte_pktmbuf_data_len(loc->mbuf); + if (MLX5_TXOFF_CONFIG(VLAN) && + loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { + vlan = sizeof(struct rte_vlan_hdr); + } + /* + * First calculate the WQE size to check + * whether we have enough space in ring buffer. + */ + hlen = loc->mbuf->l2_len + vlan + + loc->mbuf->l3_len + loc->mbuf->l4_len; + if (unlikely((!hlen || !loc->mbuf->tso_segsz))) + return MLX5_TXCMP_CODE_ERROR; + if (loc->mbuf->ol_flags & PKT_TX_TUNNEL_MASK) + hlen += loc->mbuf->outer_l2_len + + loc->mbuf->outer_l3_len; + /* Segment must contain all TSO headers. */ + if (unlikely(hlen > MLX5_MAX_TSO_HEADER || + hlen <= MLX5_ESEG_MIN_INLINE_SIZE || + hlen > (dlen + vlan))) + return MLX5_TXCMP_CODE_ERROR; + /* + * Check whether there are enough free WQEBBs: + * - Control Segment + * - Ethernet Segment + * - First Segment of inlined Ethernet data + * - ... data continued ... + * - Finishing Data Segment of pointer type + */ + ds = 4 + (hlen - MLX5_ESEG_MIN_INLINE_SIZE + + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; + if (loc->wqe_free < ((ds + 3) / 4)) + return MLX5_TXCMP_CODE_EXIT; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes/packets counters. */ + ntcp = (dlen + vlan - hlen + + loc->mbuf->tso_segsz - 1) / + loc->mbuf->tso_segsz; + /* + * One will be added for mbuf itself at the end + * of the mlx5_tx_burst from loc->pkts_sent field. + */ + --ntcp; + txq->stats.opackets += ntcp; + txq->stats.obytes += dlen + vlan + ntcp * hlen; +#endif + /* + * Build the TSO WQE: + * - Control Segment + * - Ethernet Segment with hlen bytes inlined + * - Data Segment of pointer type + */ + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, ds, + MLX5_OPCODE_TSO, olx); + dseg = mlx5_tx_eseg_data(txq, loc, wqe, vlan, hlen, 1, olx); + dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *) + hlen - vlan; + dlen -= hlen - vlan; + mlx5_tx_dseg_ptr(txq, loc, dseg, dptr, dlen, olx); + /* + * WQE is built, update the loop parameters + * and go to the next packet. + */ + txq->wqe_ci += (ds + 3) / 4; + loc->wqe_free -= (ds + 3) / 4; + if (MLX5_TXOFF_CONFIG(INLINE)) + txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; + --loc->elts_free; + ++loc->pkts_sent; + --pkts_n; + if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + loc->mbuf = *pkts++; + if (pkts_n > 1) + rte_prefetch0(*pkts); + if (MLX5_TXOFF_CONFIG(MULTI) && + unlikely(NB_SEGS(loc->mbuf) > 1)) + return MLX5_TXCMP_CODE_MULTI; + if (likely(!(loc->mbuf->ol_flags & PKT_TX_TCP_SEG))) + return MLX5_TXCMP_CODE_SINGLE; + /* Continue with the next TSO packet. */ + } + MLX5_ASSERT(false); +} + +/** + * Analyze the packet and select the best method to send. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * @param newp + * The predefined flag whether do complete check for + * multi-segment packets and TSO. + * + * @return + * MLX5_TXCMP_CODE_MULTI - multi-segment packet encountered. + * MLX5_TXCMP_CODE_TSO - TSO required, use TSO/LSO. + * MLX5_TXCMP_CODE_SINGLE - single-segment packet, use SEND. + * MLX5_TXCMP_CODE_EMPW - single-segment packet, use MPW. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_able_to_empw(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx, + bool newp) +{ + /* Check for multi-segment packet. */ + if (newp && + MLX5_TXOFF_CONFIG(MULTI) && + unlikely(NB_SEGS(loc->mbuf) > 1)) + return MLX5_TXCMP_CODE_MULTI; + /* Check for TSO packet. */ + if (newp && + MLX5_TXOFF_CONFIG(TSO) && + unlikely(loc->mbuf->ol_flags & PKT_TX_TCP_SEG)) + return MLX5_TXCMP_CODE_TSO; + /* Check if eMPW is enabled at all. */ + if (!MLX5_TXOFF_CONFIG(EMPW)) + return MLX5_TXCMP_CODE_SINGLE; + /* Check if eMPW can be engaged. */ + if (MLX5_TXOFF_CONFIG(VLAN) && + unlikely(loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) && + (!MLX5_TXOFF_CONFIG(INLINE) || + unlikely((rte_pktmbuf_data_len(loc->mbuf) + + sizeof(struct rte_vlan_hdr)) > txq->inlen_empw))) { + /* + * eMPW does not support VLAN insertion offload, we have to + * inline the entire packet but packet is too long for inlining. + */ + return MLX5_TXCMP_CODE_SINGLE; + } + return MLX5_TXCMP_CODE_EMPW; +} + +/** + * Check the next packet attributes to match with the eMPW batch ones. + * In addition, for legacy MPW the packet length is checked either. + * + * @param txq + * Pointer to TX queue structure. + * @param es + * Pointer to Ethernet Segment of eMPW batch. + * @param loc + * Pointer to burst routine local context. + * @param dlen + * Length of previous packet in MPW descriptor. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * true - packet match with eMPW batch attributes. + * false - no match, eMPW should be restarted. + */ +static __rte_always_inline bool +mlx5_tx_match_empw(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_wqe_eseg *__rte_restrict es, + struct mlx5_txq_local *__rte_restrict loc, + uint32_t dlen, + unsigned int olx) +{ + uint8_t swp_flags = 0; + + /* Compare the checksum flags, if any. */ + if (MLX5_TXOFF_CONFIG(CSUM) && + txq_ol_cksum_to_cs(loc->mbuf) != es->cs_flags) + return false; + /* Compare the Software Parser offsets and flags. */ + if (MLX5_TXOFF_CONFIG(SWP) && + (es->swp_offs != txq_mbuf_to_swp(loc, &swp_flags, olx) || + es->swp_flags != swp_flags)) + return false; + /* Fill metadata field if needed. */ + if (MLX5_TXOFF_CONFIG(METADATA) && + es->metadata != (loc->mbuf->ol_flags & PKT_TX_DYNF_METADATA ? + *RTE_FLOW_DYNF_METADATA(loc->mbuf) : 0)) + return false; + /* Legacy MPW can send packets with the same length only. */ + if (MLX5_TXOFF_CONFIG(MPW) && + dlen != rte_pktmbuf_data_len(loc->mbuf)) + return false; + /* There must be no VLAN packets in eMPW loop. */ + if (MLX5_TXOFF_CONFIG(VLAN)) + MLX5_ASSERT(!(loc->mbuf->ol_flags & PKT_TX_VLAN_PKT)); + /* Check if the scheduling is requested. */ + if (MLX5_TXOFF_CONFIG(TXPP) && + loc->mbuf->ol_flags & txq->ts_mask) + return false; + return true; +} + +/** + * Update send loop variables and WQE for eMPW loop without data inlining. + * Number of Data Segments is equal to the number of sent packets. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param ds + * Number of packets/Data Segments/Packets. + * @param slen + * Accumulated statistics, bytes sent. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * true - packet match with eMPW batch attributes. + * false - no match, eMPW should be restarted. + */ +static __rte_always_inline void +mlx5_tx_sdone_empw(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int ds, + unsigned int slen, + unsigned int olx __rte_unused) +{ + MLX5_ASSERT(!MLX5_TXOFF_CONFIG(INLINE)); +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + txq->stats.obytes += slen; +#else + (void)slen; +#endif + loc->elts_free -= ds; + loc->pkts_sent += ds; + ds += 2; + loc->wqe_last->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | ds); + txq->wqe_ci += (ds + 3) / 4; + loc->wqe_free -= (ds + 3) / 4; +} + +/** + * Update send loop variables and WQE for eMPW loop with data inlining. + * Gets the size of pushed descriptors and data to the WQE. + * + * @param txq + * Pointer to TX queue structure. + * @param loc + * Pointer to burst routine local context. + * @param len + * Total size of descriptor/data in bytes. + * @param slen + * Accumulated statistics, data bytes sent. + * @param wqem + * The base WQE for the eMPW/MPW descriptor. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * true - packet match with eMPW batch attributes. + * false - no match, eMPW should be restarted. + */ +static __rte_always_inline void +mlx5_tx_idone_empw(struct mlx5_txq_data *__rte_restrict txq, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int len, + unsigned int slen, + struct mlx5_wqe *__rte_restrict wqem, + unsigned int olx __rte_unused) +{ + struct mlx5_wqe_dseg *dseg = &wqem->dseg[0]; + + MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + txq->stats.obytes += slen; +#else + (void)slen; +#endif + if (MLX5_TXOFF_CONFIG(MPW) && dseg->bcount == RTE_BE32(0)) { + /* + * If the legacy MPW session contains the inline packets + * we should set the only inline data segment length + * and align the total length to the segment size. + */ + MLX5_ASSERT(len > sizeof(dseg->bcount)); + dseg->bcount = rte_cpu_to_be_32((len - sizeof(dseg->bcount)) | + MLX5_ETH_WQE_DATA_INLINE); + len = (len + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE + 2; + } else { + /* + * The session is not legacy MPW or contains the + * data buffer pointer segments. + */ + MLX5_ASSERT((len % MLX5_WSEG_SIZE) == 0); + len = len / MLX5_WSEG_SIZE + 2; + } + wqem->cseg.sq_ds = rte_cpu_to_be_32(txq->qp_num_8s | len); + txq->wqe_ci += (len + 3) / 4; + loc->wqe_free -= (len + 3) / 4; + loc->wqe_last = wqem; +} + +/** + * The set of Tx burst functions for single-segment packets without TSO + * and with Multi-Packet Writing feature support. + * Supports all types of Tx offloads, except multi-packets and TSO. + * + * Uses MLX5_OPCODE_EMPW to build WQEs if possible and sends as many packet + * per WQE as it can. If eMPW is not configured or packet can not be sent with + * eMPW (VLAN insertion) the ordinary SEND opcode is used and only one packet + * placed in WQE. + * + * Functions stop sending if it encounters the multi-segment packet or packet + * with TSO requested. + * + * The routines are responsible for storing processed mbuf into elts ring buffer + * and update elts_head if inlining offload is requested. Otherwise the copying + * mbufs to elts can be postponed and completed at the end of burst routine. + * + * @param txq + * Pointer to TX queue structure. + * @param[in] pkts + * Packets to transmit. + * @param pkts_n + * Number of packets in array. + * @param loc + * Pointer to burst routine local context. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * @return + * MLX5_TXCMP_CODE_EXIT - sending is done or impossible. + * MLX5_TXCMP_CODE_ERROR - some unrecoverable error occurred. + * MLX5_TXCMP_CODE_MULTI - multi-segment packet encountered. + * MLX5_TXCMP_CODE_TSO - TSO packet encountered. + * MLX5_TXCMP_CODE_SINGLE - used inside functions set. + * MLX5_TXCMP_CODE_EMPW - used inside functions set. + * + * Local context variables updated. + * + * + * The routine sends packets with MLX5_OPCODE_EMPW + * without inlining, this is dedicated optimized branch. + * No VLAN insertion is supported. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_burst_empw_simple(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + /* + * Subroutine is the part of mlx5_tx_burst_single() and sends + * single-segment packet with eMPW opcode without data inlining. + */ + MLX5_ASSERT(!MLX5_TXOFF_CONFIG(INLINE)); + MLX5_ASSERT(MLX5_TXOFF_CONFIG(EMPW)); + MLX5_ASSERT(loc->elts_free && loc->wqe_free); + MLX5_ASSERT(pkts_n > loc->pkts_sent); + pkts += loc->pkts_sent + 1; + pkts_n -= loc->pkts_sent; + for (;;) { + struct mlx5_wqe_dseg *__rte_restrict dseg; + struct mlx5_wqe_eseg *__rte_restrict eseg; + enum mlx5_txcmp_code ret; + unsigned int part, loop; + unsigned int slen = 0; + +next_empw: + MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + part = RTE_MIN(pkts_n, MLX5_TXOFF_CONFIG(MPW) ? + MLX5_MPW_MAX_PACKETS : + MLX5_EMPW_MAX_PACKETS); + if (unlikely(loc->elts_free < part)) { + /* We have no enough elts to save all mbufs. */ + if (unlikely(loc->elts_free < MLX5_EMPW_MIN_PACKETS)) + return MLX5_TXCMP_CODE_EXIT; + /* But we still able to send at least minimal eMPW. */ + part = loc->elts_free; + } + /* Check whether we have enough WQEs */ + if (unlikely(loc->wqe_free < ((2 + part + 3) / 4))) { + if (unlikely(loc->wqe_free < + ((2 + MLX5_EMPW_MIN_PACKETS + 3) / 4))) + return MLX5_TXCMP_CODE_EXIT; + part = (loc->wqe_free * 4) - 2; + } + if (likely(part > 1)) + rte_prefetch0(*pkts); + loc->wqe_last = txq->wqes + (txq->wqe_ci & txq->wqe_m); + /* + * Build eMPW title WQEBB: + * - Control Segment, eMPW opcode + * - Ethernet Segment, no inline + */ + mlx5_tx_cseg_init(txq, loc, loc->wqe_last, part + 2, + MLX5_OPCODE_ENHANCED_MPSW, olx); + mlx5_tx_eseg_none(txq, loc, loc->wqe_last, + olx & ~MLX5_TXOFF_CONFIG_VLAN); + eseg = &loc->wqe_last->eseg; + dseg = &loc->wqe_last->dseg[0]; + loop = part; + /* Store the packet length for legacy MPW. */ + if (MLX5_TXOFF_CONFIG(MPW)) + eseg->mss = rte_cpu_to_be_16 + (rte_pktmbuf_data_len(loc->mbuf)); + for (;;) { + uint32_t dlen = rte_pktmbuf_data_len(loc->mbuf); +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + slen += dlen; +#endif + mlx5_tx_dseg_ptr + (txq, loc, dseg, + rte_pktmbuf_mtod(loc->mbuf, uint8_t *), + dlen, olx); + if (unlikely(--loop == 0)) + break; + loc->mbuf = *pkts++; + if (likely(loop > 1)) + rte_prefetch0(*pkts); + ret = mlx5_tx_able_to_empw(txq, loc, olx, true); + /* + * Unroll the completion code to avoid + * returning variable value - it results in + * unoptimized sequent checking in caller. + */ + if (ret == MLX5_TXCMP_CODE_MULTI) { + part -= loop; + mlx5_tx_sdone_empw(txq, loc, part, slen, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + return MLX5_TXCMP_CODE_MULTI; + } + MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); + if (ret == MLX5_TXCMP_CODE_TSO) { + part -= loop; + mlx5_tx_sdone_empw(txq, loc, part, slen, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + return MLX5_TXCMP_CODE_TSO; + } + if (ret == MLX5_TXCMP_CODE_SINGLE) { + part -= loop; + mlx5_tx_sdone_empw(txq, loc, part, slen, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + return MLX5_TXCMP_CODE_SINGLE; + } + if (ret != MLX5_TXCMP_CODE_EMPW) { + MLX5_ASSERT(false); + part -= loop; + mlx5_tx_sdone_empw(txq, loc, part, slen, olx); + return MLX5_TXCMP_CODE_ERROR; + } + /* + * Check whether packet parameters coincide + * within assumed eMPW batch: + * - check sum settings + * - metadata value + * - software parser settings + * - packets length (legacy MPW only) + * - scheduling is not required + */ + if (!mlx5_tx_match_empw(txq, eseg, loc, dlen, olx)) { + MLX5_ASSERT(loop); + part -= loop; + mlx5_tx_sdone_empw(txq, loc, part, slen, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + pkts_n -= part; + goto next_empw; + } + /* Packet attributes match, continue the same eMPW. */ + ++dseg; + if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) + dseg = (struct mlx5_wqe_dseg *)txq->wqes; + } + /* eMPW is built successfully, update loop parameters. */ + MLX5_ASSERT(!loop); + MLX5_ASSERT(pkts_n >= part); +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + txq->stats.obytes += slen; +#endif + loc->elts_free -= part; + loc->pkts_sent += part; + txq->wqe_ci += (2 + part + 3) / 4; + loc->wqe_free -= (2 + part + 3) / 4; + pkts_n -= part; + if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + loc->mbuf = *pkts++; + ret = mlx5_tx_able_to_empw(txq, loc, olx, true); + if (unlikely(ret != MLX5_TXCMP_CODE_EMPW)) + return ret; + /* Continue sending eMPW batches. */ + } + MLX5_ASSERT(false); +} + +/** + * The routine sends packets with MLX5_OPCODE_EMPW + * with inlining, optionally supports VLAN insertion. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_burst_empw_inline(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + /* + * Subroutine is the part of mlx5_tx_burst_single() and sends + * single-segment packet with eMPW opcode with data inlining. + */ + MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); + MLX5_ASSERT(MLX5_TXOFF_CONFIG(EMPW)); + MLX5_ASSERT(loc->elts_free && loc->wqe_free); + MLX5_ASSERT(pkts_n > loc->pkts_sent); + pkts += loc->pkts_sent + 1; + pkts_n -= loc->pkts_sent; + for (;;) { + struct mlx5_wqe_dseg *__rte_restrict dseg; + struct mlx5_wqe *__rte_restrict wqem; + enum mlx5_txcmp_code ret; + unsigned int room, part, nlim; + unsigned int slen = 0; + + MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + /* + * Limits the amount of packets in one WQE + * to improve CQE latency generation. + */ + nlim = RTE_MIN(pkts_n, MLX5_TXOFF_CONFIG(MPW) ? + MLX5_MPW_INLINE_MAX_PACKETS : + MLX5_EMPW_MAX_PACKETS); + /* Check whether we have minimal amount WQEs */ + if (unlikely(loc->wqe_free < + ((2 + MLX5_EMPW_MIN_PACKETS + 3) / 4))) + return MLX5_TXCMP_CODE_EXIT; + if (likely(pkts_n > 1)) + rte_prefetch0(*pkts); + wqem = txq->wqes + (txq->wqe_ci & txq->wqe_m); + /* + * Build eMPW title WQEBB: + * - Control Segment, eMPW opcode, zero DS + * - Ethernet Segment, no inline + */ + mlx5_tx_cseg_init(txq, loc, wqem, 0, + MLX5_OPCODE_ENHANCED_MPSW, olx); + mlx5_tx_eseg_none(txq, loc, wqem, + olx & ~MLX5_TXOFF_CONFIG_VLAN); + dseg = &wqem->dseg[0]; + /* Store the packet length for legacy MPW. */ + if (MLX5_TXOFF_CONFIG(MPW)) + wqem->eseg.mss = rte_cpu_to_be_16 + (rte_pktmbuf_data_len(loc->mbuf)); + room = RTE_MIN(MLX5_WQE_SIZE_MAX / MLX5_WQE_SIZE, + loc->wqe_free) * MLX5_WQE_SIZE - + MLX5_WQE_CSEG_SIZE - + MLX5_WQE_ESEG_SIZE; + /* Limit the room for legacy MPW sessions for performance. */ + if (MLX5_TXOFF_CONFIG(MPW)) + room = RTE_MIN(room, + RTE_MAX(txq->inlen_empw + + sizeof(dseg->bcount) + + (MLX5_TXOFF_CONFIG(VLAN) ? + sizeof(struct rte_vlan_hdr) : 0), + MLX5_MPW_INLINE_MAX_PACKETS * + MLX5_WQE_DSEG_SIZE)); + /* Build WQE till we have space, packets and resources. */ + part = room; + for (;;) { + uint32_t dlen = rte_pktmbuf_data_len(loc->mbuf); + uint8_t *dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *); + unsigned int tlen; + + MLX5_ASSERT(room >= MLX5_WQE_DSEG_SIZE); + MLX5_ASSERT((room % MLX5_WQE_DSEG_SIZE) == 0); + MLX5_ASSERT((uintptr_t)dseg < (uintptr_t)txq->wqes_end); + /* + * Some Tx offloads may cause an error if packet is not + * long enough, check against assumed minimal length. + */ + if (unlikely(dlen <= MLX5_ESEG_MIN_INLINE_SIZE)) { + part -= room; + if (unlikely(!part)) + return MLX5_TXCMP_CODE_ERROR; + /* + * We have some successfully built + * packet Data Segments to send. + */ + mlx5_tx_idone_empw(txq, loc, part, + slen, wqem, olx); + return MLX5_TXCMP_CODE_ERROR; + } + /* Inline or not inline - that's the Question. */ + if (dlen > txq->inlen_empw || + loc->mbuf->ol_flags & PKT_TX_DYNF_NOINLINE) + goto pointer_empw; + if (MLX5_TXOFF_CONFIG(MPW)) { + if (dlen > txq->inlen_send) + goto pointer_empw; + tlen = dlen; + if (part == room) { + /* Open new inline MPW session. */ + tlen += sizeof(dseg->bcount); + dseg->bcount = RTE_BE32(0); + dseg = RTE_PTR_ADD + (dseg, sizeof(dseg->bcount)); + } else { + /* + * No pointer and inline descriptor + * intermix for legacy MPW sessions. + */ + if (wqem->dseg[0].bcount) + break; + } + } else { + tlen = sizeof(dseg->bcount) + dlen; + } + /* Inline entire packet, optional VLAN insertion. */ + if (MLX5_TXOFF_CONFIG(VLAN) && + loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { + /* + * The packet length must be checked in + * mlx5_tx_able_to_empw() and packet + * fits into inline length guaranteed. + */ + MLX5_ASSERT((dlen + + sizeof(struct rte_vlan_hdr)) <= + txq->inlen_empw); + tlen += sizeof(struct rte_vlan_hdr); + if (room < tlen) + break; + dseg = mlx5_tx_dseg_vlan(txq, loc, dseg, + dptr, dlen, olx); +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + slen += sizeof(struct rte_vlan_hdr); +#endif + } else { + if (room < tlen) + break; + dseg = mlx5_tx_dseg_empw(txq, loc, dseg, + dptr, dlen, olx); + } + if (!MLX5_TXOFF_CONFIG(MPW)) + tlen = RTE_ALIGN(tlen, MLX5_WSEG_SIZE); + MLX5_ASSERT(room >= tlen); + room -= tlen; + /* + * Packet data are completely inline, + * we can try to free the packet. + */ + if (likely(loc->pkts_sent == loc->mbuf_free)) { + /* + * All the packets from the burst beginning + * are inline, we can free mbufs directly + * from the origin array on tx_burst exit(). + */ + loc->mbuf_free++; + goto next_mbuf; + } + /* + * In order no to call rte_pktmbuf_free_seg() here, + * in the most inner loop (that might be very + * expensive) we just save the mbuf in elts. + */ + txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; + loc->elts_free--; + goto next_mbuf; +pointer_empw: + /* + * No pointer and inline descriptor + * intermix for legacy MPW sessions. + */ + if (MLX5_TXOFF_CONFIG(MPW) && + part != room && + wqem->dseg[0].bcount == RTE_BE32(0)) + break; + /* + * Not inlinable VLAN packets are + * proceeded outside of this routine. + */ + MLX5_ASSERT(room >= MLX5_WQE_DSEG_SIZE); + if (MLX5_TXOFF_CONFIG(VLAN)) + MLX5_ASSERT(!(loc->mbuf->ol_flags & + PKT_TX_VLAN_PKT)); + mlx5_tx_dseg_ptr(txq, loc, dseg, dptr, dlen, olx); + /* We have to store mbuf in elts.*/ + txq->elts[txq->elts_head++ & txq->elts_m] = loc->mbuf; + loc->elts_free--; + room -= MLX5_WQE_DSEG_SIZE; + /* Ring buffer wraparound is checked at the loop end.*/ + ++dseg; +next_mbuf: +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + slen += dlen; +#endif + loc->pkts_sent++; + pkts_n--; + if (unlikely(!pkts_n || !loc->elts_free)) { + /* + * We have no resources/packets to + * continue build descriptors. + */ + part -= room; + mlx5_tx_idone_empw(txq, loc, part, + slen, wqem, olx); + return MLX5_TXCMP_CODE_EXIT; + } + loc->mbuf = *pkts++; + if (likely(pkts_n > 1)) + rte_prefetch0(*pkts); + ret = mlx5_tx_able_to_empw(txq, loc, olx, true); + /* + * Unroll the completion code to avoid + * returning variable value - it results in + * unoptimized sequent checking in caller. + */ + if (ret == MLX5_TXCMP_CODE_MULTI) { + part -= room; + mlx5_tx_idone_empw(txq, loc, part, + slen, wqem, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + return MLX5_TXCMP_CODE_MULTI; + } + MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); + if (ret == MLX5_TXCMP_CODE_TSO) { + part -= room; + mlx5_tx_idone_empw(txq, loc, part, + slen, wqem, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + return MLX5_TXCMP_CODE_TSO; + } + if (ret == MLX5_TXCMP_CODE_SINGLE) { + part -= room; + mlx5_tx_idone_empw(txq, loc, part, + slen, wqem, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + return MLX5_TXCMP_CODE_SINGLE; + } + if (ret != MLX5_TXCMP_CODE_EMPW) { + MLX5_ASSERT(false); + part -= room; + mlx5_tx_idone_empw(txq, loc, part, + slen, wqem, olx); + return MLX5_TXCMP_CODE_ERROR; + } + /* Check if we have minimal room left. */ + nlim--; + if (unlikely(!nlim || room < MLX5_WQE_DSEG_SIZE)) + break; + /* + * Check whether packet parameters coincide + * within assumed eMPW batch: + * - check sum settings + * - metadata value + * - software parser settings + * - packets length (legacy MPW only) + * - scheduling is not required + */ + if (!mlx5_tx_match_empw(txq, &wqem->eseg, + loc, dlen, olx)) + break; + /* Packet attributes match, continue the same eMPW. */ + if ((uintptr_t)dseg >= (uintptr_t)txq->wqes_end) + dseg = (struct mlx5_wqe_dseg *)txq->wqes; + } + /* + * We get here to close an existing eMPW + * session and start the new one. + */ + MLX5_ASSERT(pkts_n); + part -= room; + if (unlikely(!part)) + return MLX5_TXCMP_CODE_EXIT; + mlx5_tx_idone_empw(txq, loc, part, slen, wqem, olx); + if (unlikely(!loc->elts_free || + !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + /* Continue the loop with new eMPW session. */ + } + MLX5_ASSERT(false); +} + +/** + * The routine sends packets with ordinary MLX5_OPCODE_SEND. + * Data inlining and VLAN insertion are supported. + */ +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_burst_single_send(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + /* + * Subroutine is the part of mlx5_tx_burst_single() + * and sends single-segment packet with SEND opcode. + */ + MLX5_ASSERT(loc->elts_free && loc->wqe_free); + MLX5_ASSERT(pkts_n > loc->pkts_sent); + pkts += loc->pkts_sent + 1; + pkts_n -= loc->pkts_sent; + for (;;) { + struct mlx5_wqe *__rte_restrict wqe; + enum mlx5_txcmp_code ret; + + MLX5_ASSERT(NB_SEGS(loc->mbuf) == 1); + if (MLX5_TXOFF_CONFIG(TXPP)) { + enum mlx5_txcmp_code wret; + + /* Generate WAIT for scheduling if requested. */ + wret = mlx5_tx_schedule_send(txq, loc, olx); + if (wret == MLX5_TXCMP_CODE_EXIT) + return MLX5_TXCMP_CODE_EXIT; + if (wret == MLX5_TXCMP_CODE_ERROR) + return MLX5_TXCMP_CODE_ERROR; + } + if (MLX5_TXOFF_CONFIG(INLINE)) { + unsigned int inlen, vlan = 0; + + inlen = rte_pktmbuf_data_len(loc->mbuf); + if (MLX5_TXOFF_CONFIG(VLAN) && + loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) { + vlan = sizeof(struct rte_vlan_hdr); + inlen += vlan; + } + /* + * If inlining is enabled at configuration time + * the limit must be not less than minimal size. + * Otherwise we would do extra check for data + * size to avoid crashes due to length overflow. + */ + MLX5_ASSERT(txq->inlen_send >= + MLX5_ESEG_MIN_INLINE_SIZE); + if (inlen <= txq->inlen_send) { + unsigned int seg_n, wqe_n; + + rte_prefetch0(rte_pktmbuf_mtod + (loc->mbuf, uint8_t *)); + /* Check against minimal length. */ + if (inlen <= MLX5_ESEG_MIN_INLINE_SIZE) + return MLX5_TXCMP_CODE_ERROR; + if (loc->mbuf->ol_flags & + PKT_TX_DYNF_NOINLINE) { + /* + * The hint flag not to inline packet + * data is set. Check whether we can + * follow the hint. + */ + if ((!MLX5_TXOFF_CONFIG(EMPW) && + txq->inlen_mode) || + (MLX5_TXOFF_CONFIG(MPW) && + txq->inlen_mode)) { + if (inlen <= txq->inlen_send) + goto single_inline; + /* + * The hardware requires the + * minimal inline data header. + */ + goto single_min_inline; + } + if (MLX5_TXOFF_CONFIG(VLAN) && + vlan && !txq->vlan_en) { + /* + * We must insert VLAN tag + * by software means. + */ + goto single_part_inline; + } + goto single_no_inline; + } +single_inline: + /* + * Completely inlined packet data WQE: + * - Control Segment, SEND opcode + * - Ethernet Segment, no VLAN insertion + * - Data inlined, VLAN optionally inserted + * - Alignment to MLX5_WSEG_SIZE + * Have to estimate amount of WQEBBs + */ + seg_n = (inlen + 3 * MLX5_WSEG_SIZE - + MLX5_ESEG_MIN_INLINE_SIZE + + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; + /* Check if there are enough WQEBBs. */ + wqe_n = (seg_n + 3) / 4; + if (wqe_n > loc->wqe_free) + return MLX5_TXCMP_CODE_EXIT; + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, seg_n, + MLX5_OPCODE_SEND, olx); + mlx5_tx_eseg_data(txq, loc, wqe, + vlan, inlen, 0, olx); + txq->wqe_ci += wqe_n; + loc->wqe_free -= wqe_n; + /* + * Packet data are completely inlined, + * free the packet immediately. + */ + rte_pktmbuf_free_seg(loc->mbuf); + } else if ((!MLX5_TXOFF_CONFIG(EMPW) || + MLX5_TXOFF_CONFIG(MPW)) && + txq->inlen_mode) { + /* + * If minimal inlining is requested the eMPW + * feature should be disabled due to data is + * inlined into Ethernet Segment, which can + * not contain inlined data for eMPW due to + * segment shared for all packets. + */ + struct mlx5_wqe_dseg *__rte_restrict dseg; + unsigned int ds; + uint8_t *dptr; + + /* + * The inline-mode settings require + * to inline the specified amount of + * data bytes to the Ethernet Segment. + * We should check the free space in + * WQE ring buffer to inline partially. + */ +single_min_inline: + MLX5_ASSERT(txq->inlen_send >= txq->inlen_mode); + MLX5_ASSERT(inlen > txq->inlen_mode); + MLX5_ASSERT(txq->inlen_mode >= + MLX5_ESEG_MIN_INLINE_SIZE); + /* + * Check whether there are enough free WQEBBs: + * - Control Segment + * - Ethernet Segment + * - First Segment of inlined Ethernet data + * - ... data continued ... + * - Finishing Data Segment of pointer type + */ + ds = (MLX5_WQE_CSEG_SIZE + + MLX5_WQE_ESEG_SIZE + + MLX5_WQE_DSEG_SIZE + + txq->inlen_mode - + MLX5_ESEG_MIN_INLINE_SIZE + + MLX5_WQE_DSEG_SIZE + + MLX5_WSEG_SIZE - 1) / MLX5_WSEG_SIZE; + if (loc->wqe_free < ((ds + 3) / 4)) + return MLX5_TXCMP_CODE_EXIT; + /* + * Build the ordinary SEND WQE: + * - Control Segment + * - Ethernet Segment, inline inlen_mode bytes + * - Data Segment of pointer type + */ + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, ds, + MLX5_OPCODE_SEND, olx); + dseg = mlx5_tx_eseg_data(txq, loc, wqe, vlan, + txq->inlen_mode, + 0, olx); + dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *) + + txq->inlen_mode - vlan; + inlen -= txq->inlen_mode; + mlx5_tx_dseg_ptr(txq, loc, dseg, + dptr, inlen, olx); + /* + * WQE is built, update the loop parameters + * and got to the next packet. + */ + txq->wqe_ci += (ds + 3) / 4; + loc->wqe_free -= (ds + 3) / 4; + /* We have to store mbuf in elts.*/ + MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); + txq->elts[txq->elts_head++ & txq->elts_m] = + loc->mbuf; + --loc->elts_free; + } else { + uint8_t *dptr; + unsigned int dlen; + + /* + * Partially inlined packet data WQE, we have + * some space in title WQEBB, we can fill it + * with some packet data. It takes one WQEBB, + * it is available, no extra space check: + * - Control Segment, SEND opcode + * - Ethernet Segment, no VLAN insertion + * - MLX5_ESEG_MIN_INLINE_SIZE bytes of Data + * - Data Segment, pointer type + * + * We also get here if VLAN insertion is not + * supported by HW, the inline is enabled. + */ +single_part_inline: + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, 4, + MLX5_OPCODE_SEND, olx); + mlx5_tx_eseg_dmin(txq, loc, wqe, vlan, olx); + dptr = rte_pktmbuf_mtod(loc->mbuf, uint8_t *) + + MLX5_ESEG_MIN_INLINE_SIZE - vlan; + /* + * The length check is performed above, by + * comparing with txq->inlen_send. We should + * not get overflow here. + */ + MLX5_ASSERT(inlen > MLX5_ESEG_MIN_INLINE_SIZE); + dlen = inlen - MLX5_ESEG_MIN_INLINE_SIZE; + mlx5_tx_dseg_ptr(txq, loc, &wqe->dseg[1], + dptr, dlen, olx); + ++txq->wqe_ci; + --loc->wqe_free; + /* We have to store mbuf in elts.*/ + MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE)); + txq->elts[txq->elts_head++ & txq->elts_m] = + loc->mbuf; + --loc->elts_free; + } +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + txq->stats.obytes += vlan + + rte_pktmbuf_data_len(loc->mbuf); +#endif + } else { + /* + * No inline at all, it means the CPU cycles saving + * is prioritized at configuration, we should not + * copy any packet data to WQE. + * + * SEND WQE, one WQEBB: + * - Control Segment, SEND opcode + * - Ethernet Segment, optional VLAN, no inline + * - Data Segment, pointer type + */ +single_no_inline: + wqe = txq->wqes + (txq->wqe_ci & txq->wqe_m); + loc->wqe_last = wqe; + mlx5_tx_cseg_init(txq, loc, wqe, 3, + MLX5_OPCODE_SEND, olx); + mlx5_tx_eseg_none(txq, loc, wqe, olx); + mlx5_tx_dseg_ptr + (txq, loc, &wqe->dseg[0], + rte_pktmbuf_mtod(loc->mbuf, uint8_t *), + rte_pktmbuf_data_len(loc->mbuf), olx); + ++txq->wqe_ci; + --loc->wqe_free; + /* + * We should not store mbuf pointer in elts + * if no inlining is configured, this is done + * by calling routine in a batch copy. + */ + MLX5_ASSERT(!MLX5_TXOFF_CONFIG(INLINE)); + --loc->elts_free; +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Update sent data bytes counter. */ + txq->stats.obytes += rte_pktmbuf_data_len(loc->mbuf); + if (MLX5_TXOFF_CONFIG(VLAN) && + loc->mbuf->ol_flags & PKT_TX_VLAN_PKT) + txq->stats.obytes += + sizeof(struct rte_vlan_hdr); +#endif + } + ++loc->pkts_sent; + --pkts_n; + if (unlikely(!pkts_n || !loc->elts_free || !loc->wqe_free)) + return MLX5_TXCMP_CODE_EXIT; + loc->mbuf = *pkts++; + if (pkts_n > 1) + rte_prefetch0(*pkts); + ret = mlx5_tx_able_to_empw(txq, loc, olx, true); + if (unlikely(ret != MLX5_TXCMP_CODE_SINGLE)) + return ret; + } + MLX5_ASSERT(false); +} + +static __rte_always_inline enum mlx5_txcmp_code +mlx5_tx_burst_single(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + unsigned int pkts_n, + struct mlx5_txq_local *__rte_restrict loc, + unsigned int olx) +{ + enum mlx5_txcmp_code ret; + + ret = mlx5_tx_able_to_empw(txq, loc, olx, false); + if (ret == MLX5_TXCMP_CODE_SINGLE) + goto ordinary_send; + MLX5_ASSERT(ret == MLX5_TXCMP_CODE_EMPW); + for (;;) { + /* Optimize for inline/no inline eMPW send. */ + ret = (MLX5_TXOFF_CONFIG(INLINE)) ? + mlx5_tx_burst_empw_inline + (txq, pkts, pkts_n, loc, olx) : + mlx5_tx_burst_empw_simple + (txq, pkts, pkts_n, loc, olx); + if (ret != MLX5_TXCMP_CODE_SINGLE) + return ret; + /* The resources to send one packet should remain. */ + MLX5_ASSERT(loc->elts_free && loc->wqe_free); +ordinary_send: + ret = mlx5_tx_burst_single_send(txq, pkts, pkts_n, loc, olx); + MLX5_ASSERT(ret != MLX5_TXCMP_CODE_SINGLE); + if (ret != MLX5_TXCMP_CODE_EMPW) + return ret; + /* The resources to send one packet should remain. */ + MLX5_ASSERT(loc->elts_free && loc->wqe_free); + } +} + +/** + * DPDK Tx callback template. This is configured template used to generate + * routines optimized for specified offload setup. + * One of this generated functions is chosen at SQ configuration time. + * + * @param txq + * Generic pointer to TX queue structure. + * @param[in] pkts + * Packets to transmit. + * @param pkts_n + * Number of packets in array. + * @param olx + * Configured offloads mask, presents the bits of MLX5_TXOFF_CONFIG_xxx + * values. Should be static to take compile time static configuration + * advantages. + * + * @return + * Number of packets successfully transmitted (<= pkts_n). + */ +static __rte_always_inline uint16_t +mlx5_tx_burst_tmpl(struct mlx5_txq_data *__rte_restrict txq, + struct rte_mbuf **__rte_restrict pkts, + uint16_t pkts_n, + unsigned int olx) +{ + struct mlx5_txq_local loc; + enum mlx5_txcmp_code ret; + unsigned int part; + + MLX5_ASSERT(txq->elts_s >= (uint16_t)(txq->elts_head - txq->elts_tail)); + MLX5_ASSERT(txq->wqe_s >= (uint16_t)(txq->wqe_ci - txq->wqe_pi)); + if (unlikely(!pkts_n)) + return 0; + if (MLX5_TXOFF_CONFIG(INLINE)) + loc.mbuf_free = 0; + loc.pkts_sent = 0; + loc.pkts_copy = 0; + loc.wqe_last = NULL; + +send_loop: + loc.pkts_loop = loc.pkts_sent; + /* + * Check if there are some CQEs, if any: + * - process an encountered errors + * - process the completed WQEs + * - free related mbufs + * - doorbell the NIC about processed CQEs + */ + rte_prefetch0(*(pkts + loc.pkts_sent)); + mlx5_tx_handle_completion(txq, olx); + /* + * Calculate the number of available resources - elts and WQEs. + * There are two possible different scenarios: + * - no data inlining into WQEs, one WQEBB may contains up to + * four packets, in this case elts become scarce resource + * - data inlining into WQEs, one packet may require multiple + * WQEBBs, the WQEs become the limiting factor. + */ + MLX5_ASSERT(txq->elts_s >= (uint16_t)(txq->elts_head - txq->elts_tail)); + loc.elts_free = txq->elts_s - + (uint16_t)(txq->elts_head - txq->elts_tail); + MLX5_ASSERT(txq->wqe_s >= (uint16_t)(txq->wqe_ci - txq->wqe_pi)); + loc.wqe_free = txq->wqe_s - + (uint16_t)(txq->wqe_ci - txq->wqe_pi); + if (unlikely(!loc.elts_free || !loc.wqe_free)) + goto burst_exit; + for (;;) { + /* + * Fetch the packet from array. Usually this is the first + * packet in series of multi/single segment packets. + */ + loc.mbuf = *(pkts + loc.pkts_sent); + /* Dedicated branch for multi-segment packets. */ + if (MLX5_TXOFF_CONFIG(MULTI) && + unlikely(NB_SEGS(loc.mbuf) > 1)) { + /* + * Multi-segment packet encountered. + * Hardware is able to process it only + * with SEND/TSO opcodes, one packet + * per WQE, do it in dedicated routine. + */ +enter_send_multi: + MLX5_ASSERT(loc.pkts_sent >= loc.pkts_copy); + part = loc.pkts_sent - loc.pkts_copy; + if (!MLX5_TXOFF_CONFIG(INLINE) && part) { + /* + * There are some single-segment mbufs not + * stored in elts. The mbufs must be in the + * same order as WQEs, so we must copy the + * mbufs to elts here, before the coming + * multi-segment packet mbufs is appended. + */ + mlx5_tx_copy_elts(txq, pkts + loc.pkts_copy, + part, olx); + loc.pkts_copy = loc.pkts_sent; + } + MLX5_ASSERT(pkts_n > loc.pkts_sent); + ret = mlx5_tx_burst_mseg(txq, pkts, pkts_n, &loc, olx); + if (!MLX5_TXOFF_CONFIG(INLINE)) + loc.pkts_copy = loc.pkts_sent; + /* + * These returned code checks are supposed + * to be optimized out due to routine inlining. + */ + if (ret == MLX5_TXCMP_CODE_EXIT) { + /* + * The routine returns this code when + * all packets are sent or there is no + * enough resources to complete request. + */ + break; + } + if (ret == MLX5_TXCMP_CODE_ERROR) { + /* + * The routine returns this code when some error + * in the incoming packets format occurred. + */ + txq->stats.oerrors++; + break; + } + if (ret == MLX5_TXCMP_CODE_SINGLE) { + /* + * The single-segment packet was encountered + * in the array, try to send it with the + * best optimized way, possible engaging eMPW. + */ + goto enter_send_single; + } + if (MLX5_TXOFF_CONFIG(TSO) && + ret == MLX5_TXCMP_CODE_TSO) { + /* + * The single-segment TSO packet was + * encountered in the array. + */ + goto enter_send_tso; + } + /* We must not get here. Something is going wrong. */ + MLX5_ASSERT(false); + txq->stats.oerrors++; + break; + } + /* Dedicated branch for single-segment TSO packets. */ + if (MLX5_TXOFF_CONFIG(TSO) && + unlikely(loc.mbuf->ol_flags & PKT_TX_TCP_SEG)) { + /* + * TSO might require special way for inlining + * (dedicated parameters) and is sent with + * MLX5_OPCODE_TSO opcode only, provide this + * in dedicated branch. + */ +enter_send_tso: + MLX5_ASSERT(NB_SEGS(loc.mbuf) == 1); + MLX5_ASSERT(pkts_n > loc.pkts_sent); + ret = mlx5_tx_burst_tso(txq, pkts, pkts_n, &loc, olx); + /* + * These returned code checks are supposed + * to be optimized out due to routine inlining. + */ + if (ret == MLX5_TXCMP_CODE_EXIT) + break; + if (ret == MLX5_TXCMP_CODE_ERROR) { + txq->stats.oerrors++; + break; + } + if (ret == MLX5_TXCMP_CODE_SINGLE) + goto enter_send_single; + if (MLX5_TXOFF_CONFIG(MULTI) && + ret == MLX5_TXCMP_CODE_MULTI) { + /* + * The multi-segment packet was + * encountered in the array. + */ + goto enter_send_multi; + } + /* We must not get here. Something is going wrong. */ + MLX5_ASSERT(false); + txq->stats.oerrors++; + break; + } + /* + * The dedicated branch for the single-segment packets + * without TSO. Often these ones can be sent using + * MLX5_OPCODE_EMPW with multiple packets in one WQE. + * The routine builds the WQEs till it encounters + * the TSO or multi-segment packet (in case if these + * offloads are requested at SQ configuration time). + */ +enter_send_single: + MLX5_ASSERT(pkts_n > loc.pkts_sent); + ret = mlx5_tx_burst_single(txq, pkts, pkts_n, &loc, olx); + /* + * These returned code checks are supposed + * to be optimized out due to routine inlining. + */ + if (ret == MLX5_TXCMP_CODE_EXIT) + break; + if (ret == MLX5_TXCMP_CODE_ERROR) { + txq->stats.oerrors++; + break; + } + if (MLX5_TXOFF_CONFIG(MULTI) && + ret == MLX5_TXCMP_CODE_MULTI) { + /* + * The multi-segment packet was + * encountered in the array. + */ + goto enter_send_multi; + } + if (MLX5_TXOFF_CONFIG(TSO) && + ret == MLX5_TXCMP_CODE_TSO) { + /* + * The single-segment TSO packet was + * encountered in the array. + */ + goto enter_send_tso; + } + /* We must not get here. Something is going wrong. */ + MLX5_ASSERT(false); + txq->stats.oerrors++; + break; + } + /* + * Main Tx loop is completed, do the rest: + * - set completion request if thresholds are reached + * - doorbell the hardware + * - copy the rest of mbufs to elts (if any) + */ + MLX5_ASSERT(MLX5_TXOFF_CONFIG(INLINE) || + loc.pkts_sent >= loc.pkts_copy); + /* Take a shortcut if nothing is sent. */ + if (unlikely(loc.pkts_sent == loc.pkts_loop)) + goto burst_exit; + /* Request CQE generation if limits are reached. */ + mlx5_tx_request_completion(txq, &loc, olx); + /* + * Ring QP doorbell immediately after WQE building completion + * to improve latencies. The pure software related data treatment + * can be completed after doorbell. Tx CQEs for this SQ are + * processed in this thread only by the polling. + * + * The rdma core library can map doorbell register in two ways, + * depending on the environment variable "MLX5_SHUT_UP_BF": + * + * - as regular cached memory, the variable is either missing or + * set to zero. This type of mapping may cause the significant + * doorbell register writing latency and requires explicit memory + * write barrier to mitigate this issue and prevent write combining. + * + * - as non-cached memory, the variable is present and set to not "0" + * value. This type of mapping may cause performance impact under + * heavy loading conditions but the explicit write memory barrier is + * not required and it may improve core performance. + * + * - the legacy behaviour (prior 19.08 release) was to use some + * heuristics to decide whether write memory barrier should + * be performed. This behavior is supported with specifying + * tx_db_nc=2, write barrier is skipped if application provides + * the full recommended burst of packets, it supposes the next + * packets are coming and the write barrier will be issued on + * the next burst (after descriptor writing, at least). + */ + mlx5_tx_dbrec_cond_wmb(txq, loc.wqe_last, !txq->db_nc && + (!txq->db_heu || pkts_n % MLX5_TX_DEFAULT_BURST)); + /* Not all of the mbufs may be stored into elts yet. */ + part = MLX5_TXOFF_CONFIG(INLINE) ? 0 : loc.pkts_sent - loc.pkts_copy; + if (!MLX5_TXOFF_CONFIG(INLINE) && part) { + /* + * There are some single-segment mbufs not stored in elts. + * It can be only if the last packet was single-segment. + * The copying is gathered into one place due to it is + * a good opportunity to optimize that with SIMD. + * Unfortunately if inlining is enabled the gaps in pointer + * array may happen due to early freeing of the inlined mbufs. + */ + mlx5_tx_copy_elts(txq, pkts + loc.pkts_copy, part, olx); + loc.pkts_copy = loc.pkts_sent; + } + MLX5_ASSERT(txq->elts_s >= (uint16_t)(txq->elts_head - txq->elts_tail)); + MLX5_ASSERT(txq->wqe_s >= (uint16_t)(txq->wqe_ci - txq->wqe_pi)); + if (pkts_n > loc.pkts_sent) { + /* + * If burst size is large there might be no enough CQE + * fetched from completion queue and no enough resources + * freed to send all the packets. + */ + goto send_loop; + } +burst_exit: +#ifdef MLX5_PMD_SOFT_COUNTERS + /* Increment sent packets counter. */ + txq->stats.opackets += loc.pkts_sent; +#endif + if (MLX5_TXOFF_CONFIG(INLINE) && loc.mbuf_free) + __mlx5_tx_free_mbuf(txq, pkts, loc.mbuf_free, olx); + return loc.pkts_sent; +} + #endif /* RTE_PMD_MLX5_TX_H_ */ From patchwork Mon Apr 5 14:00:55 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Baum X-Patchwork-Id: 90563 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id ECC04A0A02; Mon, 5 Apr 2021 16:02:24 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id B6D09140F2E; Mon, 5 Apr 2021 16:01:42 +0200 (CEST) Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by mails.dpdk.org (Postfix) with ESMTP id A7FA1140EC5 for ; Mon, 5 Apr 2021 16:01:35 +0200 (CEST) Received: from Internal Mail-Server by MTLPINE1 (envelope-from michaelba@nvidia.com) with SMTP; 5 Apr 2021 17:01:32 +0300 Received: from nvidia.com (pegasus07.mtr.labs.mlnx [10.210.16.112]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 135E1VNb031476; Mon, 5 Apr 2021 17:01:32 +0300 From: Michael Baum To: dev@dpdk.org Cc: Matan Azrad , Raslan Darawsheh , Viacheslav Ovsiienko Date: Mon, 5 Apr 2021 14:00:55 +0000 Message-Id: <1617631256-3018-6-git-send-email-michaelba@nvidia.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> References: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> Subject: [dpdk-dev] [PATCH 5/6] net/mlx5: separate Tx function implementations to new file X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" This patch separates Tx function implementations to different source file as an optional preparation step for Tx cleanup. Signed-off-by: Michael Baum Acked-by: Viacheslav Ovsiienko --- drivers/net/mlx5/meson.build | 1 + drivers/net/mlx5/mlx5_rxtx.c | 757 ----------------------------------------- drivers/net/mlx5/mlx5_tx.c | 780 +++++++++++++++++++++++++++++++++++++++++++ drivers/net/mlx5/mlx5_tx.h | 59 +++- drivers/net/mlx5/mlx5_txq.c | 1 + 5 files changed, 838 insertions(+), 760 deletions(-) create mode 100644 drivers/net/mlx5/mlx5_tx.c diff --git a/drivers/net/mlx5/meson.build b/drivers/net/mlx5/meson.build index 0a89a27..688a925 100644 --- a/drivers/net/mlx5/meson.build +++ b/drivers/net/mlx5/meson.build @@ -25,6 +25,7 @@ sources = files( 'mlx5_rxtx.c', 'mlx5_stats.c', 'mlx5_trigger.c', + 'mlx5_tx.c', 'mlx5_txq.c', 'mlx5_txpp.c', 'mlx5_vlan.c', diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c index 2f36754..35c4cc3 100644 --- a/drivers/net/mlx5/mlx5_rxtx.c +++ b/drivers/net/mlx5/mlx5_rxtx.c @@ -28,8 +28,6 @@ #include "mlx5_rx.h" #include "mlx5_tx.h" -#define MLX5_TXOFF_INFO(func, olx) {mlx5_tx_burst_##func, olx}, - /* static asserts */ static_assert(MLX5_CQE_STATUS_HW_OWN < 0, "Must be negative value"); static_assert(MLX5_CQE_STATUS_SW_OWN < 0, "Must be negative value"); @@ -357,113 +355,6 @@ } /** - * Move QP from error state to running state and initialize indexes. - * - * @param txq_ctrl - * Pointer to TX queue control structure. - * - * @return - * 0 on success, else -1. - */ -static int -tx_recover_qp(struct mlx5_txq_ctrl *txq_ctrl) -{ - struct mlx5_mp_arg_queue_state_modify sm = { - .is_wq = 0, - .queue_id = txq_ctrl->txq.idx, - }; - - if (mlx5_queue_state_modify(ETH_DEV(txq_ctrl->priv), &sm)) - return -1; - txq_ctrl->txq.wqe_ci = 0; - txq_ctrl->txq.wqe_pi = 0; - txq_ctrl->txq.elts_comp = 0; - return 0; -} - -/* Return 1 if the error CQE is signed otherwise, sign it and return 0. */ -static int -check_err_cqe_seen(volatile struct mlx5_err_cqe *err_cqe) -{ - static const uint8_t magic[] = "seen"; - int ret = 1; - unsigned int i; - - for (i = 0; i < sizeof(magic); ++i) - if (!ret || err_cqe->rsvd1[i] != magic[i]) { - ret = 0; - err_cqe->rsvd1[i] = magic[i]; - } - return ret; -} - -/** - * Handle error CQE. - * - * @param txq - * Pointer to TX queue structure. - * @param error_cqe - * Pointer to the error CQE. - * - * @return - * Negative value if queue recovery failed, otherwise - * the error completion entry is handled successfully. - */ -static int -mlx5_tx_error_cqe_handle(struct mlx5_txq_data *__rte_restrict txq, - volatile struct mlx5_err_cqe *err_cqe) -{ - if (err_cqe->syndrome != MLX5_CQE_SYNDROME_WR_FLUSH_ERR) { - const uint16_t wqe_m = ((1 << txq->wqe_n) - 1); - struct mlx5_txq_ctrl *txq_ctrl = - container_of(txq, struct mlx5_txq_ctrl, txq); - uint16_t new_wqe_pi = rte_be_to_cpu_16(err_cqe->wqe_counter); - int seen = check_err_cqe_seen(err_cqe); - - if (!seen && txq_ctrl->dump_file_n < - txq_ctrl->priv->config.max_dump_files_num) { - MKSTR(err_str, "Unexpected CQE error syndrome " - "0x%02x CQN = %u SQN = %u wqe_counter = %u " - "wq_ci = %u cq_ci = %u", err_cqe->syndrome, - txq->cqe_s, txq->qp_num_8s >> 8, - rte_be_to_cpu_16(err_cqe->wqe_counter), - txq->wqe_ci, txq->cq_ci); - MKSTR(name, "dpdk_mlx5_port_%u_txq_%u_index_%u_%u", - PORT_ID(txq_ctrl->priv), txq->idx, - txq_ctrl->dump_file_n, (uint32_t)rte_rdtsc()); - mlx5_dump_debug_information(name, NULL, err_str, 0); - mlx5_dump_debug_information(name, "MLX5 Error CQ:", - (const void *)((uintptr_t) - txq->cqes), - sizeof(*err_cqe) * - (1 << txq->cqe_n)); - mlx5_dump_debug_information(name, "MLX5 Error SQ:", - (const void *)((uintptr_t) - txq->wqes), - MLX5_WQE_SIZE * - (1 << txq->wqe_n)); - txq_ctrl->dump_file_n++; - } - if (!seen) - /* - * Count errors in WQEs units. - * Later it can be improved to count error packets, - * for example, by SQ parsing to find how much packets - * should be counted for each WQE. - */ - txq->stats.oerrors += ((txq->wqe_ci & wqe_m) - - new_wqe_pi) & wqe_m; - if (tx_recover_qp(txq_ctrl)) { - /* Recovering failed - retry later on the same WQE. */ - return -1; - } - /* Release all the remaining buffers. */ - txq_free_elts(txq_ctrl); - } - return 0; -} - -/** * Modify a Verbs/DevX queue state. * This must be called from the primary process. * @@ -539,174 +430,6 @@ return ret; } -/** - * Dummy DPDK callback for TX. - * - * This function is used to temporarily replace the real callback during - * unsafe control operations on the queue, or in case of error. - * - * @param dpdk_txq - * Generic pointer to TX queue structure. - * @param[in] pkts - * Packets to transmit. - * @param pkts_n - * Number of packets in array. - * - * @return - * Number of packets successfully transmitted (<= pkts_n). - */ -uint16_t -removed_tx_burst(void *dpdk_txq __rte_unused, - struct rte_mbuf **pkts __rte_unused, - uint16_t pkts_n __rte_unused) -{ - rte_mb(); - return 0; -} - -/** - * Update completion queue consuming index via doorbell - * and flush the completed data buffers. - * - * @param txq - * Pointer to TX queue structure. - * @param valid CQE pointer - * if not NULL update txq->wqe_pi and flush the buffers - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - */ -static __rte_always_inline void -mlx5_tx_comp_flush(struct mlx5_txq_data *__rte_restrict txq, - volatile struct mlx5_cqe *last_cqe, - unsigned int olx __rte_unused) -{ - if (likely(last_cqe != NULL)) { - uint16_t tail; - - txq->wqe_pi = rte_be_to_cpu_16(last_cqe->wqe_counter); - tail = txq->fcqs[(txq->cq_ci - 1) & txq->cqe_m]; - if (likely(tail != txq->elts_tail)) { - mlx5_tx_free_elts(txq, tail, olx); - MLX5_ASSERT(tail == txq->elts_tail); - } - } -} - -/** - * Manage TX completions. This routine checks the CQ for - * arrived CQEs, deduces the last accomplished WQE in SQ, - * updates SQ producing index and frees all completed mbufs. - * - * @param txq - * Pointer to TX queue structure. - * @param olx - * Configured Tx offloads mask. It is fully defined at - * compile time and may be used for optimization. - * - * NOTE: not inlined intentionally, it makes tx_burst - * routine smaller, simple and faster - from experiments. - */ -void -mlx5_tx_handle_completion(struct mlx5_txq_data *__rte_restrict txq, - unsigned int olx __rte_unused) -{ - unsigned int count = MLX5_TX_COMP_MAX_CQE; - volatile struct mlx5_cqe *last_cqe = NULL; - bool ring_doorbell = false; - int ret; - - do { - volatile struct mlx5_cqe *cqe; - - cqe = &txq->cqes[txq->cq_ci & txq->cqe_m]; - ret = check_cqe(cqe, txq->cqe_s, txq->cq_ci); - if (unlikely(ret != MLX5_CQE_STATUS_SW_OWN)) { - if (likely(ret != MLX5_CQE_STATUS_ERR)) { - /* No new CQEs in completion queue. */ - MLX5_ASSERT(ret == MLX5_CQE_STATUS_HW_OWN); - break; - } - /* - * Some error occurred, try to restart. - * We have no barrier after WQE related Doorbell - * written, make sure all writes are completed - * here, before we might perform SQ reset. - */ - rte_wmb(); - ret = mlx5_tx_error_cqe_handle - (txq, (volatile struct mlx5_err_cqe *)cqe); - if (unlikely(ret < 0)) { - /* - * Some error occurred on queue error - * handling, we do not advance the index - * here, allowing to retry on next call. - */ - return; - } - /* - * We are going to fetch all entries with - * MLX5_CQE_SYNDROME_WR_FLUSH_ERR status. - * The send queue is supposed to be empty. - */ - ring_doorbell = true; - ++txq->cq_ci; - txq->cq_pi = txq->cq_ci; - last_cqe = NULL; - continue; - } - /* Normal transmit completion. */ - MLX5_ASSERT(txq->cq_ci != txq->cq_pi); -#ifdef RTE_LIBRTE_MLX5_DEBUG - MLX5_ASSERT((txq->fcqs[txq->cq_ci & txq->cqe_m] >> 16) == - cqe->wqe_counter); -#endif - ring_doorbell = true; - ++txq->cq_ci; - last_cqe = cqe; - /* - * We have to restrict the amount of processed CQEs - * in one tx_burst routine call. The CQ may be large - * and many CQEs may be updated by the NIC in one - * transaction. Buffers freeing is time consuming, - * multiple iterations may introduce significant - * latency. - */ - if (likely(--count == 0)) - break; - } while (true); - if (likely(ring_doorbell)) { - /* Ring doorbell to notify hardware. */ - rte_compiler_barrier(); - *txq->cq_db = rte_cpu_to_be_32(txq->cq_ci); - mlx5_tx_comp_flush(txq, last_cqe, olx); - } -} - -/** - * DPDK callback to check the status of a tx descriptor. - * - * @param tx_queue - * The tx queue. - * @param[in] offset - * The index of the descriptor in the ring. - * - * @return - * The status of the tx descriptor. - */ -int -mlx5_tx_descriptor_status(void *tx_queue, uint16_t offset) -{ - struct mlx5_txq_data *__rte_restrict txq = tx_queue; - uint16_t used; - - mlx5_tx_handle_completion(txq, 0); - used = txq->elts_head - txq->elts_tail; - if (offset < used) - return RTE_ETH_TX_DESC_FULL; - return RTE_ETH_TX_DESC_DONE; -} - /* Generate routines with Enhanced Multi-Packet Write support. */ MLX5_TXOFF_DECL(full_empw, MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_EMPW) @@ -907,483 +630,3 @@ MLX5_TXOFF_DECL(i_mpw, MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | MLX5_TXOFF_CONFIG_MPW) - -/* - * Array of declared and compiled Tx burst function and corresponding - * supported offloads set. The array is used to select the Tx burst - * function for specified offloads set at Tx queue configuration time. - */ -const struct { - eth_tx_burst_t func; - unsigned int olx; -} txoff_func[] = { -MLX5_TXOFF_INFO(full_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(none_empw, - MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(md_empw, - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mt_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mtsc_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mti_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mtv_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mtiv_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(sc_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(sci_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(scv_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(sciv_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(i_empw, - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(v_empw, - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(iv_empw, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(full_ts_nompw, - MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP) - -MLX5_TXOFF_INFO(full_ts_nompwi, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP) - -MLX5_TXOFF_INFO(full_ts, - MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP | - MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(full_ts_noi, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(none_ts, - MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_TXPP | - MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mdi_ts, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mti_ts, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(mtiv_ts, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_TXPP | - MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_INFO(full, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(none, - MLX5_TXOFF_CONFIG_NONE) - -MLX5_TXOFF_INFO(md, - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(mt, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(mtsc, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(mti, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(mtv, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(mtiv, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(sc, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(sci, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(scv, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(sciv, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(i, - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(v, - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(iv, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_INFO(none_mpw, - MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW) - -MLX5_TXOFF_INFO(mci_mpw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW) - -MLX5_TXOFF_INFO(mc_mpw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_EMPW | MLX5_TXOFF_CONFIG_MPW) - -MLX5_TXOFF_INFO(i_mpw, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW) -}; - -/** - * Configure the Tx function to use. The routine checks configured - * Tx offloads for the device and selects appropriate Tx burst - * routine. There are multiple Tx burst routines compiled from - * the same template in the most optimal way for the dedicated - * Tx offloads set. - * - * @param dev - * Pointer to private data structure. - * - * @return - * Pointer to selected Tx burst function. - */ -eth_tx_burst_t -mlx5_select_tx_function(struct rte_eth_dev *dev) -{ - struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_dev_config *config = &priv->config; - uint64_t tx_offloads = dev->data->dev_conf.txmode.offloads; - unsigned int diff = 0, olx = 0, i, m; - - MLX5_ASSERT(priv); - if (tx_offloads & DEV_TX_OFFLOAD_MULTI_SEGS) { - /* We should support Multi-Segment Packets. */ - olx |= MLX5_TXOFF_CONFIG_MULTI; - } - if (tx_offloads & (DEV_TX_OFFLOAD_TCP_TSO | - DEV_TX_OFFLOAD_VXLAN_TNL_TSO | - DEV_TX_OFFLOAD_GRE_TNL_TSO | - DEV_TX_OFFLOAD_IP_TNL_TSO | - DEV_TX_OFFLOAD_UDP_TNL_TSO)) { - /* We should support TCP Send Offload. */ - olx |= MLX5_TXOFF_CONFIG_TSO; - } - if (tx_offloads & (DEV_TX_OFFLOAD_IP_TNL_TSO | - DEV_TX_OFFLOAD_UDP_TNL_TSO | - DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM)) { - /* We should support Software Parser for Tunnels. */ - olx |= MLX5_TXOFF_CONFIG_SWP; - } - if (tx_offloads & (DEV_TX_OFFLOAD_IPV4_CKSUM | - DEV_TX_OFFLOAD_UDP_CKSUM | - DEV_TX_OFFLOAD_TCP_CKSUM | - DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM)) { - /* We should support IP/TCP/UDP Checksums. */ - olx |= MLX5_TXOFF_CONFIG_CSUM; - } - if (tx_offloads & DEV_TX_OFFLOAD_VLAN_INSERT) { - /* We should support VLAN insertion. */ - olx |= MLX5_TXOFF_CONFIG_VLAN; - } - if (tx_offloads & DEV_TX_OFFLOAD_SEND_ON_TIMESTAMP && - rte_mbuf_dynflag_lookup - (RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME, NULL) >= 0 && - rte_mbuf_dynfield_lookup - (RTE_MBUF_DYNFIELD_TIMESTAMP_NAME, NULL) >= 0) { - /* Offload configured, dynamic entities registered. */ - olx |= MLX5_TXOFF_CONFIG_TXPP; - } - if (priv->txqs_n && (*priv->txqs)[0]) { - struct mlx5_txq_data *txd = (*priv->txqs)[0]; - - if (txd->inlen_send) { - /* - * Check the data inline requirements. Data inline - * is enabled on per device basis, we can check - * the first Tx queue only. - * - * If device does not support VLAN insertion in WQE - * and some queues are requested to perform VLAN - * insertion offload than inline must be enabled. - */ - olx |= MLX5_TXOFF_CONFIG_INLINE; - } - } - if (config->mps == MLX5_MPW_ENHANCED && - config->txq_inline_min <= 0) { - /* - * The NIC supports Enhanced Multi-Packet Write - * and does not require minimal inline data. - */ - olx |= MLX5_TXOFF_CONFIG_EMPW; - } - if (rte_flow_dynf_metadata_avail()) { - /* We should support Flow metadata. */ - olx |= MLX5_TXOFF_CONFIG_METADATA; - } - if (config->mps == MLX5_MPW) { - /* - * The NIC supports Legacy Multi-Packet Write. - * The MLX5_TXOFF_CONFIG_MPW controls the - * descriptor building method in combination - * with MLX5_TXOFF_CONFIG_EMPW. - */ - if (!(olx & (MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA))) - olx |= MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW; - } - /* - * Scan the routines table to find the minimal - * satisfying routine with requested offloads. - */ - m = RTE_DIM(txoff_func); - for (i = 0; i < RTE_DIM(txoff_func); i++) { - unsigned int tmp; - - tmp = txoff_func[i].olx; - if (tmp == olx) { - /* Meets requested offloads exactly.*/ - m = i; - break; - } - if ((tmp & olx) != olx) { - /* Does not meet requested offloads at all. */ - continue; - } - if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_MPW) - /* Do not enable legacy MPW if not configured. */ - continue; - if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_EMPW) - /* Do not enable eMPW if not configured. */ - continue; - if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_INLINE) - /* Do not enable inlining if not configured. */ - continue; - if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_TXPP) - /* Do not enable scheduling if not configured. */ - continue; - /* - * Some routine meets the requirements. - * Check whether it has minimal amount - * of not requested offloads. - */ - tmp = __builtin_popcountl(tmp & ~olx); - if (m >= RTE_DIM(txoff_func) || tmp < diff) { - /* First or better match, save and continue. */ - m = i; - diff = tmp; - continue; - } - if (tmp == diff) { - tmp = txoff_func[i].olx ^ txoff_func[m].olx; - if (__builtin_ffsl(txoff_func[i].olx & ~tmp) < - __builtin_ffsl(txoff_func[m].olx & ~tmp)) { - /* Lighter not requested offload. */ - m = i; - } - } - } - if (m >= RTE_DIM(txoff_func)) { - DRV_LOG(DEBUG, "port %u has no selected Tx function" - " for requested offloads %04X", - dev->data->port_id, olx); - return NULL; - } - DRV_LOG(DEBUG, "port %u has selected Tx function" - " supporting offloads %04X/%04X", - dev->data->port_id, olx, txoff_func[m].olx); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_MULTI) - DRV_LOG(DEBUG, "\tMULTI (multi segment)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_TSO) - DRV_LOG(DEBUG, "\tTSO (TCP send offload)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_SWP) - DRV_LOG(DEBUG, "\tSWP (software parser)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_CSUM) - DRV_LOG(DEBUG, "\tCSUM (checksum offload)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_INLINE) - DRV_LOG(DEBUG, "\tINLIN (inline data)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_VLAN) - DRV_LOG(DEBUG, "\tVLANI (VLAN insertion)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_METADATA) - DRV_LOG(DEBUG, "\tMETAD (tx Flow metadata)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_TXPP) - DRV_LOG(DEBUG, "\tMETAD (tx Scheduling)"); - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_EMPW) { - if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_MPW) - DRV_LOG(DEBUG, "\tMPW (Legacy MPW)"); - else - DRV_LOG(DEBUG, "\tEMPW (Enhanced MPW)"); - } - return txoff_func[m].func; -} - -/** - * DPDK callback to get the TX queue information - * - * @param dev - * Pointer to the device structure. - * - * @param tx_queue_id - * Tx queue identificator. - * - * @param qinfo - * Pointer to the TX queue information structure. - * - * @return - * None. - */ - -void -mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, - struct rte_eth_txq_info *qinfo) -{ - struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_txq_data *txq = (*priv->txqs)[tx_queue_id]; - struct mlx5_txq_ctrl *txq_ctrl = - container_of(txq, struct mlx5_txq_ctrl, txq); - - if (!txq) - return; - qinfo->nb_desc = txq->elts_s; - qinfo->conf.tx_thresh.pthresh = 0; - qinfo->conf.tx_thresh.hthresh = 0; - qinfo->conf.tx_thresh.wthresh = 0; - qinfo->conf.tx_rs_thresh = 0; - qinfo->conf.tx_free_thresh = 0; - qinfo->conf.tx_deferred_start = txq_ctrl ? 0 : 1; - qinfo->conf.offloads = dev->data->dev_conf.txmode.offloads; -} - -/** - * DPDK callback to get the TX packet burst mode information - * - * @param dev - * Pointer to the device structure. - * - * @param tx_queue_id - * Tx queue identificatior. - * - * @param mode - * Pointer to the burts mode information. - * - * @return - * 0 as success, -EINVAL as failure. - */ - -int -mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, - uint16_t tx_queue_id, - struct rte_eth_burst_mode *mode) -{ - eth_tx_burst_t pkt_burst = dev->tx_pkt_burst; - struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_txq_data *txq = (*priv->txqs)[tx_queue_id]; - unsigned int i, olx; - - for (i = 0; i < RTE_DIM(txoff_func); i++) { - if (pkt_burst == txoff_func[i].func) { - olx = txoff_func[i].olx; - snprintf(mode->info, sizeof(mode->info), - "%s%s%s%s%s%s%s%s%s%s", - (olx & MLX5_TXOFF_CONFIG_EMPW) ? - ((olx & MLX5_TXOFF_CONFIG_MPW) ? - "Legacy MPW" : "Enhanced MPW") : "No MPW", - (olx & MLX5_TXOFF_CONFIG_MULTI) ? - " + MULTI" : "", - (olx & MLX5_TXOFF_CONFIG_TSO) ? - " + TSO" : "", - (olx & MLX5_TXOFF_CONFIG_SWP) ? - " + SWP" : "", - (olx & MLX5_TXOFF_CONFIG_CSUM) ? - " + CSUM" : "", - (olx & MLX5_TXOFF_CONFIG_INLINE) ? - " + INLINE" : "", - (olx & MLX5_TXOFF_CONFIG_VLAN) ? - " + VLAN" : "", - (olx & MLX5_TXOFF_CONFIG_METADATA) ? - " + METADATA" : "", - (olx & MLX5_TXOFF_CONFIG_TXPP) ? - " + TXPP" : "", - (txq && txq->fast_free) ? - " + Fast Free" : ""); - return 0; - } - } - return -EINVAL; -} diff --git a/drivers/net/mlx5/mlx5_tx.c b/drivers/net/mlx5/mlx5_tx.c new file mode 100644 index 0000000..df67137 --- /dev/null +++ b/drivers/net/mlx5/mlx5_tx.c @@ -0,0 +1,780 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "mlx5_autoconf.h" +#include "mlx5_defs.h" +#include "mlx5.h" +#include "mlx5_mr.h" +#include "mlx5_utils.h" +#include "mlx5_rxtx.h" +#include "mlx5_tx.h" + +#define MLX5_TXOFF_INFO(func, olx) {mlx5_tx_burst_##func, olx}, + +/** + * Move QP from error state to running state and initialize indexes. + * + * @param txq_ctrl + * Pointer to TX queue control structure. + * + * @return + * 0 on success, else -1. + */ +static int +tx_recover_qp(struct mlx5_txq_ctrl *txq_ctrl) +{ + struct mlx5_mp_arg_queue_state_modify sm = { + .is_wq = 0, + .queue_id = txq_ctrl->txq.idx, + }; + + if (mlx5_queue_state_modify(ETH_DEV(txq_ctrl->priv), &sm)) + return -1; + txq_ctrl->txq.wqe_ci = 0; + txq_ctrl->txq.wqe_pi = 0; + txq_ctrl->txq.elts_comp = 0; + return 0; +} + +/* Return 1 if the error CQE is signed otherwise, sign it and return 0. */ +static int +check_err_cqe_seen(volatile struct mlx5_err_cqe *err_cqe) +{ + static const uint8_t magic[] = "seen"; + int ret = 1; + unsigned int i; + + for (i = 0; i < sizeof(magic); ++i) + if (!ret || err_cqe->rsvd1[i] != magic[i]) { + ret = 0; + err_cqe->rsvd1[i] = magic[i]; + } + return ret; +} + +/** + * Handle error CQE. + * + * @param txq + * Pointer to TX queue structure. + * @param error_cqe + * Pointer to the error CQE. + * + * @return + * Negative value if queue recovery failed, otherwise + * the error completion entry is handled successfully. + */ +static int +mlx5_tx_error_cqe_handle(struct mlx5_txq_data *__rte_restrict txq, + volatile struct mlx5_err_cqe *err_cqe) +{ + if (err_cqe->syndrome != MLX5_CQE_SYNDROME_WR_FLUSH_ERR) { + const uint16_t wqe_m = ((1 << txq->wqe_n) - 1); + struct mlx5_txq_ctrl *txq_ctrl = + container_of(txq, struct mlx5_txq_ctrl, txq); + uint16_t new_wqe_pi = rte_be_to_cpu_16(err_cqe->wqe_counter); + int seen = check_err_cqe_seen(err_cqe); + + if (!seen && txq_ctrl->dump_file_n < + txq_ctrl->priv->config.max_dump_files_num) { + MKSTR(err_str, "Unexpected CQE error syndrome " + "0x%02x CQN = %u SQN = %u wqe_counter = %u " + "wq_ci = %u cq_ci = %u", err_cqe->syndrome, + txq->cqe_s, txq->qp_num_8s >> 8, + rte_be_to_cpu_16(err_cqe->wqe_counter), + txq->wqe_ci, txq->cq_ci); + MKSTR(name, "dpdk_mlx5_port_%u_txq_%u_index_%u_%u", + PORT_ID(txq_ctrl->priv), txq->idx, + txq_ctrl->dump_file_n, (uint32_t)rte_rdtsc()); + mlx5_dump_debug_information(name, NULL, err_str, 0); + mlx5_dump_debug_information(name, "MLX5 Error CQ:", + (const void *)((uintptr_t) + txq->cqes), + sizeof(*err_cqe) * + (1 << txq->cqe_n)); + mlx5_dump_debug_information(name, "MLX5 Error SQ:", + (const void *)((uintptr_t) + txq->wqes), + MLX5_WQE_SIZE * + (1 << txq->wqe_n)); + txq_ctrl->dump_file_n++; + } + if (!seen) + /* + * Count errors in WQEs units. + * Later it can be improved to count error packets, + * for example, by SQ parsing to find how much packets + * should be counted for each WQE. + */ + txq->stats.oerrors += ((txq->wqe_ci & wqe_m) - + new_wqe_pi) & wqe_m; + if (tx_recover_qp(txq_ctrl)) { + /* Recovering failed - retry later on the same WQE. */ + return -1; + } + /* Release all the remaining buffers. */ + txq_free_elts(txq_ctrl); + } + return 0; +} + +/** + * Dummy DPDK callback for TX. + * + * This function is used to temporarily replace the real callback during + * unsafe control operations on the queue, or in case of error. + * + * @param dpdk_txq + * Generic pointer to TX queue structure. + * @param[in] pkts + * Packets to transmit. + * @param pkts_n + * Number of packets in array. + * + * @return + * Number of packets successfully transmitted (<= pkts_n). + */ +uint16_t +removed_tx_burst(void *dpdk_txq __rte_unused, + struct rte_mbuf **pkts __rte_unused, + uint16_t pkts_n __rte_unused) +{ + rte_mb(); + return 0; +} + +/** + * Update completion queue consuming index via doorbell + * and flush the completed data buffers. + * + * @param txq + * Pointer to TX queue structure. + * @param last_cqe + * valid CQE pointer, if not NULL update txq->wqe_pi and flush the buffers. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + */ +static __rte_always_inline void +mlx5_tx_comp_flush(struct mlx5_txq_data *__rte_restrict txq, + volatile struct mlx5_cqe *last_cqe, + unsigned int olx __rte_unused) +{ + if (likely(last_cqe != NULL)) { + uint16_t tail; + + txq->wqe_pi = rte_be_to_cpu_16(last_cqe->wqe_counter); + tail = txq->fcqs[(txq->cq_ci - 1) & txq->cqe_m]; + if (likely(tail != txq->elts_tail)) { + mlx5_tx_free_elts(txq, tail, olx); + MLX5_ASSERT(tail == txq->elts_tail); + } + } +} + +/** + * Manage TX completions. This routine checks the CQ for + * arrived CQEs, deduces the last accomplished WQE in SQ, + * updates SQ producing index and frees all completed mbufs. + * + * @param txq + * Pointer to TX queue structure. + * @param olx + * Configured Tx offloads mask. It is fully defined at + * compile time and may be used for optimization. + * + * NOTE: not inlined intentionally, it makes tx_burst + * routine smaller, simple and faster - from experiments. + */ +void +mlx5_tx_handle_completion(struct mlx5_txq_data *__rte_restrict txq, + unsigned int olx __rte_unused) +{ + unsigned int count = MLX5_TX_COMP_MAX_CQE; + volatile struct mlx5_cqe *last_cqe = NULL; + bool ring_doorbell = false; + int ret; + + do { + volatile struct mlx5_cqe *cqe; + + cqe = &txq->cqes[txq->cq_ci & txq->cqe_m]; + ret = check_cqe(cqe, txq->cqe_s, txq->cq_ci); + if (unlikely(ret != MLX5_CQE_STATUS_SW_OWN)) { + if (likely(ret != MLX5_CQE_STATUS_ERR)) { + /* No new CQEs in completion queue. */ + MLX5_ASSERT(ret == MLX5_CQE_STATUS_HW_OWN); + break; + } + /* + * Some error occurred, try to restart. + * We have no barrier after WQE related Doorbell + * written, make sure all writes are completed + * here, before we might perform SQ reset. + */ + rte_wmb(); + ret = mlx5_tx_error_cqe_handle + (txq, (volatile struct mlx5_err_cqe *)cqe); + if (unlikely(ret < 0)) { + /* + * Some error occurred on queue error + * handling, we do not advance the index + * here, allowing to retry on next call. + */ + return; + } + /* + * We are going to fetch all entries with + * MLX5_CQE_SYNDROME_WR_FLUSH_ERR status. + * The send queue is supposed to be empty. + */ + ring_doorbell = true; + ++txq->cq_ci; + txq->cq_pi = txq->cq_ci; + last_cqe = NULL; + continue; + } + /* Normal transmit completion. */ + MLX5_ASSERT(txq->cq_ci != txq->cq_pi); +#ifdef RTE_LIBRTE_MLX5_DEBUG + MLX5_ASSERT((txq->fcqs[txq->cq_ci & txq->cqe_m] >> 16) == + cqe->wqe_counter); +#endif + ring_doorbell = true; + ++txq->cq_ci; + last_cqe = cqe; + /* + * We have to restrict the amount of processed CQEs + * in one tx_burst routine call. The CQ may be large + * and many CQEs may be updated by the NIC in one + * transaction. Buffers freeing is time consuming, + * multiple iterations may introduce significant latency. + */ + if (likely(--count == 0)) + break; + } while (true); + if (likely(ring_doorbell)) { + /* Ring doorbell to notify hardware. */ + rte_compiler_barrier(); + *txq->cq_db = rte_cpu_to_be_32(txq->cq_ci); + mlx5_tx_comp_flush(txq, last_cqe, olx); + } +} + +/** + * DPDK callback to check the status of a Tx descriptor. + * + * @param tx_queue + * The Tx queue. + * @param[in] offset + * The index of the descriptor in the ring. + * + * @return + * The status of the Tx descriptor. + */ +int +mlx5_tx_descriptor_status(void *tx_queue, uint16_t offset) +{ + struct mlx5_txq_data *__rte_restrict txq = tx_queue; + uint16_t used; + + mlx5_tx_handle_completion(txq, 0); + used = txq->elts_head - txq->elts_tail; + if (offset < used) + return RTE_ETH_TX_DESC_FULL; + return RTE_ETH_TX_DESC_DONE; +} + +/* + * Array of declared and compiled Tx burst function and corresponding + * supported offloads set. The array is used to select the Tx burst + * function for specified offloads set at Tx queue configuration time. + */ +const struct { + eth_tx_burst_t func; + unsigned int olx; +} txoff_func[] = { +MLX5_TXOFF_INFO(full_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(none_empw, + MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(md_empw, + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mt_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mtsc_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mti_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mtv_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mtiv_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(sc_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(sci_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(scv_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(sciv_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(i_empw, + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(v_empw, + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(iv_empw, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(full_ts_nompw, + MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP) + +MLX5_TXOFF_INFO(full_ts_nompwi, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP) + +MLX5_TXOFF_INFO(full_ts, + MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP | + MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(full_ts_noi, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(none_ts, + MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_TXPP | + MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mdi_ts, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mti_ts, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(mtiv_ts, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_TXPP | + MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_INFO(full, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(none, + MLX5_TXOFF_CONFIG_NONE) + +MLX5_TXOFF_INFO(md, + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(mt, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(mtsc, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(mti, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(mtv, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(mtiv, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(sc, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(sci, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(scv, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(sciv, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(i, + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(v, + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(iv, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_INFO(none_mpw, + MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW) + +MLX5_TXOFF_INFO(mci_mpw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW) + +MLX5_TXOFF_INFO(mc_mpw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_EMPW | MLX5_TXOFF_CONFIG_MPW) + +MLX5_TXOFF_INFO(i_mpw, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW) +}; + +/** + * Configure the Tx function to use. The routine checks configured + * Tx offloads for the device and selects appropriate Tx burst routine. + * There are multiple Tx burst routines compiled from the same template + * in the most optimal way for the dedicated Tx offloads set. + * + * @param dev + * Pointer to private data structure. + * + * @return + * Pointer to selected Tx burst function. + */ +eth_tx_burst_t +mlx5_select_tx_function(struct rte_eth_dev *dev) +{ + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_dev_config *config = &priv->config; + uint64_t tx_offloads = dev->data->dev_conf.txmode.offloads; + unsigned int diff = 0, olx = 0, i, m; + + MLX5_ASSERT(priv); + if (tx_offloads & DEV_TX_OFFLOAD_MULTI_SEGS) { + /* We should support Multi-Segment Packets. */ + olx |= MLX5_TXOFF_CONFIG_MULTI; + } + if (tx_offloads & (DEV_TX_OFFLOAD_TCP_TSO | + DEV_TX_OFFLOAD_VXLAN_TNL_TSO | + DEV_TX_OFFLOAD_GRE_TNL_TSO | + DEV_TX_OFFLOAD_IP_TNL_TSO | + DEV_TX_OFFLOAD_UDP_TNL_TSO)) { + /* We should support TCP Send Offload. */ + olx |= MLX5_TXOFF_CONFIG_TSO; + } + if (tx_offloads & (DEV_TX_OFFLOAD_IP_TNL_TSO | + DEV_TX_OFFLOAD_UDP_TNL_TSO | + DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM)) { + /* We should support Software Parser for Tunnels. */ + olx |= MLX5_TXOFF_CONFIG_SWP; + } + if (tx_offloads & (DEV_TX_OFFLOAD_IPV4_CKSUM | + DEV_TX_OFFLOAD_UDP_CKSUM | + DEV_TX_OFFLOAD_TCP_CKSUM | + DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM)) { + /* We should support IP/TCP/UDP Checksums. */ + olx |= MLX5_TXOFF_CONFIG_CSUM; + } + if (tx_offloads & DEV_TX_OFFLOAD_VLAN_INSERT) { + /* We should support VLAN insertion. */ + olx |= MLX5_TXOFF_CONFIG_VLAN; + } + if (tx_offloads & DEV_TX_OFFLOAD_SEND_ON_TIMESTAMP && + rte_mbuf_dynflag_lookup + (RTE_MBUF_DYNFLAG_TX_TIMESTAMP_NAME, NULL) >= 0 && + rte_mbuf_dynfield_lookup + (RTE_MBUF_DYNFIELD_TIMESTAMP_NAME, NULL) >= 0) { + /* Offload configured, dynamic entities registered. */ + olx |= MLX5_TXOFF_CONFIG_TXPP; + } + if (priv->txqs_n && (*priv->txqs)[0]) { + struct mlx5_txq_data *txd = (*priv->txqs)[0]; + + if (txd->inlen_send) { + /* + * Check the data inline requirements. Data inline + * is enabled on per device basis, we can check + * the first Tx queue only. + * + * If device does not support VLAN insertion in WQE + * and some queues are requested to perform VLAN + * insertion offload than inline must be enabled. + */ + olx |= MLX5_TXOFF_CONFIG_INLINE; + } + } + if (config->mps == MLX5_MPW_ENHANCED && + config->txq_inline_min <= 0) { + /* + * The NIC supports Enhanced Multi-Packet Write + * and does not require minimal inline data. + */ + olx |= MLX5_TXOFF_CONFIG_EMPW; + } + if (rte_flow_dynf_metadata_avail()) { + /* We should support Flow metadata. */ + olx |= MLX5_TXOFF_CONFIG_METADATA; + } + if (config->mps == MLX5_MPW) { + /* + * The NIC supports Legacy Multi-Packet Write. + * The MLX5_TXOFF_CONFIG_MPW controls the descriptor building + * method in combination with MLX5_TXOFF_CONFIG_EMPW. + */ + if (!(olx & (MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA))) + olx |= MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW; + } + /* + * Scan the routines table to find the minimal + * satisfying routine with requested offloads. + */ + m = RTE_DIM(txoff_func); + for (i = 0; i < RTE_DIM(txoff_func); i++) { + unsigned int tmp; + + tmp = txoff_func[i].olx; + if (tmp == olx) { + /* Meets requested offloads exactly.*/ + m = i; + break; + } + if ((tmp & olx) != olx) { + /* Does not meet requested offloads at all. */ + continue; + } + if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_MPW) + /* Do not enable legacy MPW if not configured. */ + continue; + if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_EMPW) + /* Do not enable eMPW if not configured. */ + continue; + if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_INLINE) + /* Do not enable inlining if not configured. */ + continue; + if ((olx ^ tmp) & MLX5_TXOFF_CONFIG_TXPP) + /* Do not enable scheduling if not configured. */ + continue; + /* + * Some routine meets the requirements. + * Check whether it has minimal amount + * of not requested offloads. + */ + tmp = __builtin_popcountl(tmp & ~olx); + if (m >= RTE_DIM(txoff_func) || tmp < diff) { + /* First or better match, save and continue. */ + m = i; + diff = tmp; + continue; + } + if (tmp == diff) { + tmp = txoff_func[i].olx ^ txoff_func[m].olx; + if (__builtin_ffsl(txoff_func[i].olx & ~tmp) < + __builtin_ffsl(txoff_func[m].olx & ~tmp)) { + /* Lighter not requested offload. */ + m = i; + } + } + } + if (m >= RTE_DIM(txoff_func)) { + DRV_LOG(DEBUG, "port %u has no selected Tx function" + " for requested offloads %04X", + dev->data->port_id, olx); + return NULL; + } + DRV_LOG(DEBUG, "port %u has selected Tx function" + " supporting offloads %04X/%04X", + dev->data->port_id, olx, txoff_func[m].olx); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_MULTI) + DRV_LOG(DEBUG, "\tMULTI (multi segment)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_TSO) + DRV_LOG(DEBUG, "\tTSO (TCP send offload)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_SWP) + DRV_LOG(DEBUG, "\tSWP (software parser)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_CSUM) + DRV_LOG(DEBUG, "\tCSUM (checksum offload)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_INLINE) + DRV_LOG(DEBUG, "\tINLIN (inline data)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_VLAN) + DRV_LOG(DEBUG, "\tVLANI (VLAN insertion)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_METADATA) + DRV_LOG(DEBUG, "\tMETAD (tx Flow metadata)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_TXPP) + DRV_LOG(DEBUG, "\tMETAD (tx Scheduling)"); + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_EMPW) { + if (txoff_func[m].olx & MLX5_TXOFF_CONFIG_MPW) + DRV_LOG(DEBUG, "\tMPW (Legacy MPW)"); + else + DRV_LOG(DEBUG, "\tEMPW (Enhanced MPW)"); + } + return txoff_func[m].func; +} + +/** + * DPDK callback to get the TX queue information. + * + * @param dev + * Pointer to the device structure. + * + * @param tx_queue_id + * Tx queue identificator. + * + * @param qinfo + * Pointer to the TX queue information structure. + * + * @return + * None. + */ +void +mlx5_txq_info_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, + struct rte_eth_txq_info *qinfo) +{ + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_txq_data *txq = (*priv->txqs)[tx_queue_id]; + struct mlx5_txq_ctrl *txq_ctrl = + container_of(txq, struct mlx5_txq_ctrl, txq); + + if (!txq) + return; + qinfo->nb_desc = txq->elts_s; + qinfo->conf.tx_thresh.pthresh = 0; + qinfo->conf.tx_thresh.hthresh = 0; + qinfo->conf.tx_thresh.wthresh = 0; + qinfo->conf.tx_rs_thresh = 0; + qinfo->conf.tx_free_thresh = 0; + qinfo->conf.tx_deferred_start = txq_ctrl ? 0 : 1; + qinfo->conf.offloads = dev->data->dev_conf.txmode.offloads; +} + +/** + * DPDK callback to get the TX packet burst mode information. + * + * @param dev + * Pointer to the device structure. + * + * @param tx_queue_id + * Tx queue identificatior. + * + * @param mode + * Pointer to the burts mode information. + * + * @return + * 0 as success, -EINVAL as failure. + */ +int +mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, + uint16_t tx_queue_id, + struct rte_eth_burst_mode *mode) +{ + eth_tx_burst_t pkt_burst = dev->tx_pkt_burst; + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_txq_data *txq = (*priv->txqs)[tx_queue_id]; + unsigned int i, olx; + + for (i = 0; i < RTE_DIM(txoff_func); i++) { + if (pkt_burst == txoff_func[i].func) { + olx = txoff_func[i].olx; + snprintf(mode->info, sizeof(mode->info), + "%s%s%s%s%s%s%s%s%s%s", + (olx & MLX5_TXOFF_CONFIG_EMPW) ? + ((olx & MLX5_TXOFF_CONFIG_MPW) ? + "Legacy MPW" : "Enhanced MPW") : "No MPW", + (olx & MLX5_TXOFF_CONFIG_MULTI) ? + " + MULTI" : "", + (olx & MLX5_TXOFF_CONFIG_TSO) ? + " + TSO" : "", + (olx & MLX5_TXOFF_CONFIG_SWP) ? + " + SWP" : "", + (olx & MLX5_TXOFF_CONFIG_CSUM) ? + " + CSUM" : "", + (olx & MLX5_TXOFF_CONFIG_INLINE) ? + " + INLINE" : "", + (olx & MLX5_TXOFF_CONFIG_VLAN) ? + " + VLAN" : "", + (olx & MLX5_TXOFF_CONFIG_METADATA) ? + " + METADATA" : "", + (olx & MLX5_TXOFF_CONFIG_TXPP) ? + " + TXPP" : "", + (txq && txq->fast_free) ? + " + Fast Free" : ""); + return 0; + } + } + return -EINVAL; +} diff --git a/drivers/net/mlx5/mlx5_tx.h b/drivers/net/mlx5/mlx5_tx.h index 34843d4..7a03aaf 100644 --- a/drivers/net/mlx5/mlx5_tx.h +++ b/drivers/net/mlx5/mlx5_tx.h @@ -62,10 +62,15 @@ enum mlx5_txcmp_code { #define MLX5_TXOFF_CONFIG(mask) (olx & MLX5_TXOFF_CONFIG_##mask) +#define MLX5_TXOFF_PRE_DECL(func) \ +uint16_t mlx5_tx_burst_##func(void *txq, \ + struct rte_mbuf **pkts, \ + uint16_t pkts_n) + #define MLX5_TXOFF_DECL(func, olx) \ -static uint16_t mlx5_tx_burst_##func(void *txq, \ - struct rte_mbuf **pkts, \ - uint16_t pkts_n) \ +uint16_t mlx5_tx_burst_##func(void *txq, \ + struct rte_mbuf **pkts, \ + uint16_t pkts_n) \ { \ return mlx5_tx_burst_tmpl((struct mlx5_txq_data *)txq, \ pkts, pkts_n, (olx)); \ @@ -237,6 +242,54 @@ int mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, struct rte_mempool *mp); +/* mlx5_rxtx.c */ + +MLX5_TXOFF_PRE_DECL(full_empw); +MLX5_TXOFF_PRE_DECL(none_empw); +MLX5_TXOFF_PRE_DECL(md_empw); +MLX5_TXOFF_PRE_DECL(mt_empw); +MLX5_TXOFF_PRE_DECL(mtsc_empw); +MLX5_TXOFF_PRE_DECL(mti_empw); +MLX5_TXOFF_PRE_DECL(mtv_empw); +MLX5_TXOFF_PRE_DECL(mtiv_empw); +MLX5_TXOFF_PRE_DECL(sc_empw); +MLX5_TXOFF_PRE_DECL(sci_empw); +MLX5_TXOFF_PRE_DECL(scv_empw); +MLX5_TXOFF_PRE_DECL(sciv_empw); +MLX5_TXOFF_PRE_DECL(i_empw); +MLX5_TXOFF_PRE_DECL(v_empw); +MLX5_TXOFF_PRE_DECL(iv_empw); + +MLX5_TXOFF_PRE_DECL(full); +MLX5_TXOFF_PRE_DECL(none); +MLX5_TXOFF_PRE_DECL(md); +MLX5_TXOFF_PRE_DECL(mt); +MLX5_TXOFF_PRE_DECL(mtsc); +MLX5_TXOFF_PRE_DECL(mti); +MLX5_TXOFF_PRE_DECL(mtv); +MLX5_TXOFF_PRE_DECL(mtiv); +MLX5_TXOFF_PRE_DECL(sc); +MLX5_TXOFF_PRE_DECL(sci); +MLX5_TXOFF_PRE_DECL(scv); +MLX5_TXOFF_PRE_DECL(sciv); +MLX5_TXOFF_PRE_DECL(i); +MLX5_TXOFF_PRE_DECL(v); +MLX5_TXOFF_PRE_DECL(iv); + +MLX5_TXOFF_PRE_DECL(full_ts_nompw); +MLX5_TXOFF_PRE_DECL(full_ts_nompwi); +MLX5_TXOFF_PRE_DECL(full_ts); +MLX5_TXOFF_PRE_DECL(full_ts_noi); +MLX5_TXOFF_PRE_DECL(none_ts); +MLX5_TXOFF_PRE_DECL(mdi_ts); +MLX5_TXOFF_PRE_DECL(mti_ts); +MLX5_TXOFF_PRE_DECL(mtiv_ts); + +MLX5_TXOFF_PRE_DECL(none_mpw); +MLX5_TXOFF_PRE_DECL(mci_mpw); +MLX5_TXOFF_PRE_DECL(mc_mpw); +MLX5_TXOFF_PRE_DECL(i_mpw); + static __rte_always_inline uint64_t * mlx5_tx_bfreg(struct mlx5_txq_data *txq) { diff --git a/drivers/net/mlx5/mlx5_txq.c b/drivers/net/mlx5/mlx5_txq.c index b8a1657..62d603e 100644 --- a/drivers/net/mlx5/mlx5_txq.c +++ b/drivers/net/mlx5/mlx5_txq.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include From patchwork Mon Apr 5 14:00:56 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Michael Baum X-Patchwork-Id: 90561 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id E604FA0A02; Mon, 5 Apr 2021 16:02:04 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 30B56140F11; Mon, 5 Apr 2021 16:01:40 +0200 (CEST) Received: from mellanox.co.il (mail-il-dmz.mellanox.com [193.47.165.129]) by mails.dpdk.org (Postfix) with ESMTP id 98D4D140E97 for ; Mon, 5 Apr 2021 16:01:35 +0200 (CEST) Received: from Internal Mail-Server by MTLPINE1 (envelope-from michaelba@nvidia.com) with SMTP; 5 Apr 2021 17:01:32 +0300 Received: from nvidia.com (pegasus07.mtr.labs.mlnx [10.210.16.112]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id 135E1VNc031476; Mon, 5 Apr 2021 17:01:32 +0300 From: Michael Baum To: dev@dpdk.org Cc: Matan Azrad , Raslan Darawsheh , Viacheslav Ovsiienko Date: Mon, 5 Apr 2021 14:00:56 +0000 Message-Id: <1617631256-3018-7-git-send-email-michaelba@nvidia.com> X-Mailer: git-send-email 1.8.3.1 In-Reply-To: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> References: <1617631256-3018-1-git-send-email-michaelba@nvidia.com> Subject: [dpdk-dev] [PATCH 6/6] net/mlx5: separate Tx burst functions to different files X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" This patch separates Tx burst function implementations to different source files, thus allowing them to compile in parallel. Signed-off-by: Michael Baum Acked-by: Viacheslav Ovsiienko --- drivers/net/mlx5/meson.build | 4 + drivers/net/mlx5/mlx5_rxtx.c | 201 --------------------------------------- drivers/net/mlx5/mlx5_tx.h | 8 +- drivers/net/mlx5/mlx5_tx_empw.c | 71 ++++++++++++++ drivers/net/mlx5/mlx5_tx_mpw.c | 34 +++++++ drivers/net/mlx5/mlx5_tx_nompw.c | 71 ++++++++++++++ drivers/net/mlx5/mlx5_tx_txpp.c | 45 +++++++++ 7 files changed, 232 insertions(+), 202 deletions(-) create mode 100644 drivers/net/mlx5/mlx5_tx_empw.c create mode 100644 drivers/net/mlx5/mlx5_tx_mpw.c create mode 100644 drivers/net/mlx5/mlx5_tx_nompw.c create mode 100644 drivers/net/mlx5/mlx5_tx_txpp.c diff --git a/drivers/net/mlx5/meson.build b/drivers/net/mlx5/meson.build index 688a925..59afd3f 100644 --- a/drivers/net/mlx5/meson.build +++ b/drivers/net/mlx5/meson.build @@ -26,6 +26,10 @@ sources = files( 'mlx5_stats.c', 'mlx5_trigger.c', 'mlx5_tx.c', + 'mlx5_tx_empw.c', + 'mlx5_tx_mpw.c', + 'mlx5_tx_nompw.c', + 'mlx5_tx_txpp.c', 'mlx5_txq.c', 'mlx5_txpp.c', 'mlx5_vlan.c', diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c index 35c4cc3..7b984ef 100644 --- a/drivers/net/mlx5/mlx5_rxtx.c +++ b/drivers/net/mlx5/mlx5_rxtx.c @@ -429,204 +429,3 @@ } return ret; } - -/* Generate routines with Enhanced Multi-Packet Write support. */ -MLX5_TXOFF_DECL(full_empw, - MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(none_empw, - MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(md_empw, - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mt_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mtsc_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mti_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mtv_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mtiv_empw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(sc_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(sci_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(scv_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(sciv_empw, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(i_empw, - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(v_empw, - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(iv_empw, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) - -/* Generate routines without Enhanced Multi-Packet Write support. */ -MLX5_TXOFF_DECL(full, - MLX5_TXOFF_CONFIG_FULL) - -MLX5_TXOFF_DECL(none, - MLX5_TXOFF_CONFIG_NONE) - -MLX5_TXOFF_DECL(md, - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(mt, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(mtsc, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(mti, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA) - - -MLX5_TXOFF_DECL(mtv, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - - -MLX5_TXOFF_DECL(mtiv, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(sc, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(sci, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA) - - -MLX5_TXOFF_DECL(scv, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - - -MLX5_TXOFF_DECL(sciv, - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(i, - MLX5_TXOFF_CONFIG_INLINE | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(v, - MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -MLX5_TXOFF_DECL(iv, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA) - -/* Generate routines with timestamp scheduling. */ -MLX5_TXOFF_DECL(full_ts_nompw, - MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP) - -MLX5_TXOFF_DECL(full_ts_nompwi, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP) - -MLX5_TXOFF_DECL(full_ts, - MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP | - MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(full_ts_noi, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(none_ts, - MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_TXPP | - MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mdi_ts, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mti_ts, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | - MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) - -MLX5_TXOFF_DECL(mtiv_ts, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | - MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_TXPP | - MLX5_TXOFF_CONFIG_EMPW) - -/* - * Generate routines with Legacy Multi-Packet Write support. - * This mode is supported by ConnectX-4 Lx only and imposes - * offload limitations, not supported: - * - ACL/Flows (metadata are becoming meaningless) - * - WQE Inline headers - * - SRIOV (E-Switch offloads) - * - VLAN insertion - * - tunnel encapsulation/decapsulation - * - TSO - */ -MLX5_TXOFF_DECL(none_mpw, - MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW) - -MLX5_TXOFF_DECL(mci_mpw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW) - -MLX5_TXOFF_DECL(mc_mpw, - MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | - MLX5_TXOFF_CONFIG_EMPW | MLX5_TXOFF_CONFIG_MPW) - -MLX5_TXOFF_DECL(i_mpw, - MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | - MLX5_TXOFF_CONFIG_MPW) diff --git a/drivers/net/mlx5/mlx5_tx.h b/drivers/net/mlx5/mlx5_tx.h index 7a03aaf..7d3ff84 100644 --- a/drivers/net/mlx5/mlx5_tx.h +++ b/drivers/net/mlx5/mlx5_tx.h @@ -242,7 +242,7 @@ int mlx5_tx_burst_mode_get(struct rte_eth_dev *dev, uint16_t tx_queue_id, uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, struct rte_mempool *mp); -/* mlx5_rxtx.c */ +/* mlx5_tx_empw.c */ MLX5_TXOFF_PRE_DECL(full_empw); MLX5_TXOFF_PRE_DECL(none_empw); @@ -260,6 +260,8 @@ uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, MLX5_TXOFF_PRE_DECL(v_empw); MLX5_TXOFF_PRE_DECL(iv_empw); +/* mlx5_tx_nompw.c */ + MLX5_TXOFF_PRE_DECL(full); MLX5_TXOFF_PRE_DECL(none); MLX5_TXOFF_PRE_DECL(md); @@ -276,6 +278,8 @@ uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, MLX5_TXOFF_PRE_DECL(v); MLX5_TXOFF_PRE_DECL(iv); +/* mlx5_tx_txpp.c */ + MLX5_TXOFF_PRE_DECL(full_ts_nompw); MLX5_TXOFF_PRE_DECL(full_ts_nompwi); MLX5_TXOFF_PRE_DECL(full_ts); @@ -285,6 +289,8 @@ uint32_t mlx5_tx_update_ext_mp(struct mlx5_txq_data *txq, uintptr_t addr, MLX5_TXOFF_PRE_DECL(mti_ts); MLX5_TXOFF_PRE_DECL(mtiv_ts); +/* mlx5_tx_mpw.c */ + MLX5_TXOFF_PRE_DECL(none_mpw); MLX5_TXOFF_PRE_DECL(mci_mpw); MLX5_TXOFF_PRE_DECL(mc_mpw); diff --git a/drivers/net/mlx5/mlx5_tx_empw.c b/drivers/net/mlx5/mlx5_tx_empw.c new file mode 100644 index 0000000..81c2dc4 --- /dev/null +++ b/drivers/net/mlx5/mlx5_tx_empw.c @@ -0,0 +1,71 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#include "mlx5_tx.h" + +/* Generate routines with Enhanced Multi-Packet Write support. */ +MLX5_TXOFF_DECL(full_empw, + MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(none_empw, + MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(md_empw, + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mt_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mtsc_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mti_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mtv_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mtiv_empw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(sc_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(sci_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(scv_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(sciv_empw, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(i_empw, + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(v_empw, + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(iv_empw, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_EMPW) diff --git a/drivers/net/mlx5/mlx5_tx_mpw.c b/drivers/net/mlx5/mlx5_tx_mpw.c new file mode 100644 index 0000000..bd7d5d2 --- /dev/null +++ b/drivers/net/mlx5/mlx5_tx_mpw.c @@ -0,0 +1,34 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#include "mlx5_tx.h" + +/* + * Generate routines with Legacy Multi-Packet Write support. + * This mode is supported by ConnectX-4 Lx only and imposes + * offload limitations, not supported: + * - ACL/Flows (metadata are becoming meaningless) + * - WQE Inline headers + * - SRIOV (E-Switch offloads) + * - VLAN insertion + * - tunnel encapsulation/decapsulation + * - TSO + */ +MLX5_TXOFF_DECL(none_mpw, + MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW) + +MLX5_TXOFF_DECL(mci_mpw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW) + +MLX5_TXOFF_DECL(mc_mpw, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_EMPW | MLX5_TXOFF_CONFIG_MPW) + +MLX5_TXOFF_DECL(i_mpw, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_EMPW | + MLX5_TXOFF_CONFIG_MPW) diff --git a/drivers/net/mlx5/mlx5_tx_nompw.c b/drivers/net/mlx5/mlx5_tx_nompw.c new file mode 100644 index 0000000..ad23aab --- /dev/null +++ b/drivers/net/mlx5/mlx5_tx_nompw.c @@ -0,0 +1,71 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#include "mlx5_tx.h" + +/* Generate routines without Enhanced Multi-Packet Write support. */ +MLX5_TXOFF_DECL(full, + MLX5_TXOFF_CONFIG_FULL) + +MLX5_TXOFF_DECL(none, + MLX5_TXOFF_CONFIG_NONE) + +MLX5_TXOFF_DECL(md, + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(mt, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(mtsc, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(mti, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(mtv, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(mtiv, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(sc, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(sci, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(scv, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(sciv, + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(i, + MLX5_TXOFF_CONFIG_INLINE | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(v, + MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) + +MLX5_TXOFF_DECL(iv, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA) diff --git a/drivers/net/mlx5/mlx5_tx_txpp.c b/drivers/net/mlx5/mlx5_tx_txpp.c new file mode 100644 index 0000000..4d2a4c6 --- /dev/null +++ b/drivers/net/mlx5/mlx5_tx_txpp.c @@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: BSD-3-Clause + * Copyright 2021 6WIND S.A. + * Copyright 2021 Mellanox Technologies, Ltd + */ + +#include "mlx5_tx.h" + +/* Generate routines with timestamp scheduling. */ +MLX5_TXOFF_DECL(full_ts_nompw, + MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP) + +MLX5_TXOFF_DECL(full_ts_nompwi, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP) + +MLX5_TXOFF_DECL(full_ts, + MLX5_TXOFF_CONFIG_FULL | MLX5_TXOFF_CONFIG_TXPP | + MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(full_ts_noi, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_SWP | MLX5_TXOFF_CONFIG_CSUM | + MLX5_TXOFF_CONFIG_VLAN | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(none_ts, + MLX5_TXOFF_CONFIG_NONE | MLX5_TXOFF_CONFIG_TXPP | + MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mdi_ts, + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mti_ts, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_METADATA | + MLX5_TXOFF_CONFIG_TXPP | MLX5_TXOFF_CONFIG_EMPW) + +MLX5_TXOFF_DECL(mtiv_ts, + MLX5_TXOFF_CONFIG_MULTI | MLX5_TXOFF_CONFIG_TSO | + MLX5_TXOFF_CONFIG_INLINE | MLX5_TXOFF_CONFIG_VLAN | + MLX5_TXOFF_CONFIG_METADATA | MLX5_TXOFF_CONFIG_TXPP | + MLX5_TXOFF_CONFIG_EMPW)